1

Onomasiology Online 6 (2005): 1-55

PAVOL ŠTEKAUER / DON CHAPMAN / SLÁVKA TOMAŠČÍKOVÁ / ŠTEFAN FRANKO

WORD-FORMATION AS CREATIVITY WITHIN PRODUCTIVITY CONSTRAINTS:SOCIOLINGUISTIC EVIDENCE

Abstract

Productivity has been one of the central topics in the field of word-formation in recent decades. Heretofore,productivity has been mainly, if not solely, discussed in formal terms, such as which affixes can be used withwhich stems, the productivity of rival affixes, etc. Such a formal approach leaves out the speakers’ needs forcreating new words. Accounting for speakers’ word-formation needs requires a re-evaluation of the notion ofcreativity. In our approach to word-formation, this notion emphasizes the active role of language users,reflecting the fact that, in each act of naming, there is more or less significant space for a coiner’s individualselection out of the options. Since each individual has unequal experiences, knowledge, intellectual capacity,imagination, education, age, professional interests, and so on, one would expect speakers to bringconsiderable variation to the naming task. Therefore, this article examines the influence of education,profession, and language-background upon the act of naming and the related word-formation productivity. Inaddition, we will examine, whether and to what degree these factors exert any influence upon the resolution ofthe fundamental conflict in word-formation (and language in general), namely that between the explicitness ofexpression and the economy of expression.

1. Introduction

Productivity has been one of the central topics in the field of word-formation in recentdecades. It was especially the 1990s and the turn of the millenium that brought new andcomprehensive insights into this field, presented by, inter alia, H. Baayen,1 I. Plag (1999),and L. Bauer (2001). Their excellent and seminal studies may be considered a culminationof a long-term effort by derivational morphologists to identify the nature of productivity inword-formation, especially with regard to the deep-rooted belief that productivity of word-formation processes is much lower than that of syntactic and inflectional processes (cf.Chomsky 1970). Gains in recent decades have shown that word-formation is moreproductive than first thought, when one is careful about the definition of productivity. Butthis paper argues that those refinements to the concept of productivity have not gone farenough. Heretofore, productivity has been mainly discussed in formal terms, such as whichaffixes can be used with which stems, the productivity of rival affixes, etc. Such a formalapproach leaves out the speakers’ needs for creating terms, and leaving out those needs hasbeen precisely what has skewed evaluation of productivity in word-formation. When thoseneeds are taken into account, word-formation seems to be as productive as syntax. It is thatclaim that this paper will argue.

Accounting for speakers’ word-formation needs requires a re-evaluation of the notion ofcreativity. Traditionally, creativity within word-formation has usually referred toidiosyncrasies and deviations from rules. In contrast, the term productivity has been usedto apply to regular, or rule-governed patterns. Word-formation theory has largely limited

The authors would like to thank Ingo Plag and an anonymous reviewer for their comments on an earlierversion of this paper.

1 A series of articles. Cf. some of them in the References.

2

itself to productivity in this sense, as it has mainly considered productivity within aframework of rival affixes (or, patterns of word-formation) used in generativemorphology, such as -ity vs. -ness. In our approach to word-formation, however, theconcept of creativity applies more generally to any act of naming by individual speakers ofa language, whether idiosyncratic or regular. Of course a speaker’s choices in the act ofnaming will be constrained by the speaker’s language system (langue), but usually, there ismore or less significant space for a coiner’s individual selection out of the options availablefor the act of naming related to a particular object of extra-linguistic reality. Since eachindividual has unequal experiences, knowledge, intellectual capacity, imagination,education, age, professional interests, and so on, one would expect speakers to bringconsiderable variation to the naming task. Thus, the notion of rivalry, more prevalent intraditional, static views of word-formation, is just a part of a much more comprehensiveconcept of the act of naming, the concept whose focal point is the active role of languageusers.

An examination of naming needs, as opposed to the distribution of formal affixes, isaccommodated within the onomasiological theory of word-formation developed byŠtekauer (1998) and his subsequent publications. In this paper, we will extend Štekauer’snotion of a Word-Formation Type cluster to cover three other levels, namely theOnomasiological Type cluster, Morphological Type cluster (see also Štekauer 2003), andWord-Formation Rule cluster. And if indeed speakers vary in their naming strategiesaccording to their different experiences, we ought to find such variation correlating withsocial variables, such as education, profession, and language background.

A further aim of this paper is to test that assumption. In particular, the integrated theory ofproductivity presented in this paper will be tested and illustrated in a questionnaire-basedevaluation of the influence of sociolinguistic factors upon acts of word-formation, and, byimplication, of the word-formation productivity. In addition, we will examine, whether andto what degree these factors exert any influence upon the resolution of the fundamentalconflict in word-formation (and language in general), that between the explicitness ofexpression and the economy of expression.2

In short, this paper aims to examine the larger notion of creativity that includes a speaker’snaming needs within an onomasiological theory of word-formation, and to demonstrate theusefulness of such an examination with a sociolinguistic study of speakers’ word-formationchoices.

The purpose of Section 1 is to provide a brief introduction to the topic. For more profoundanalyses of the state-of-the-art in the field, the reader is referred to Plag (1999), and mainlyBauer (2001).

2 We are aware of a subtler classification in cognitive linguistics, such as that proposed by Geeraerts (1983)who – at the level of what we label as ‘economy of expression’ - distinguishes between conceptualeffciency (metaphor, metonymy) and formal effciency (ellipsis, folk-etymology, avoidance of homonymicclash), and – at the level of our ‘explicitness of expression’ - between conceptual expressivity (wordformation, borrowing, semantic change) and formal expressivity (creation of specific word-formationpatterns). See also Grzega (2002: 1029ff). However, the opposition economy : explicitness fits thepurpose of our analysis.

3

1.1. At the beginning...

The beginnings were gloomy, and the outlook poor. Noam Chomsky (1970) sentencedword-formation productivity to the inferior position, with the master being the productivityof syntax and inflectional morphology. Chomsky emphasized the idiosyncratic semanticand phonological character of nominals derived from verbs, and concluded that theaccidental character of word-formation is more typical of lexical structure. According toChomsky, word-formation processes, unlike syntactic and inflectional processes, cannot beaccounted for with productive transformational rules. Not surprisingly, thetransformationalist approach to word-formation (such as Lees 1960) gave way to thelexicalist position which unambiguously separates the issues of word-formation from theissues of syntax based on the recognition that “word structure and sentence structure werenot governed by the same set of principles, and that they belonged to different modules ofthe grammar” (Mohanan 1986: 4).

But many of Chomsky’s arguments are open to objection. To Chomsky’s argument thatspecific affixes do not attach to all possible bases, two possible directions of argumentationcan be suggested (Štekauer 1998: 84ff). If we pursue the formal approach we can illustratethat the limitations on productivity operate over syntax as much as morphology (assuggested by Di Sciullo & Williams 1987), and these limitations are of the same nature. Itis true, for instance, that the suffix -ion does not combine with all verbs. But it is equallytrue that not all verbs can be used in the sentence structure N – V – Object. The limitationpermits only transitive verbs to be inserted. Both limitations (syntactic and morphological)are based on the same principle – they pertain to the combinability of structural units. Formore examples, see Di Sciullo & Williams (1987). These authors seem to have beeninspired by the following observation of S. R. Anderson:

It is true that different verbs take different formations (describe/description, laugh/laughter,recite/recital, etc.); but the point is that some action nominal formation is available for every verb(subject only to semantic limitations). One cannot really say that the diversity of the forms involvedis a limitation on the productivity of the process, any more than the existence of varying conjugationclasses constitutes a limitation on the productivity of verbal inflection in languages in which theseare found (1982: 585-586).

In this connection, S. L. Strauss also maintains that “we cannot really claim thatderivational morphology is any more idiosyncratic than the other structure-generatingrules”; in addition, “rules of derivational morphology are as regular, both semantically andphonologically, as other generative rules” (1982: 23, 24).

A second line of argumentation is of pragmatic-generative nature. If we concentrate on thegeneration aspect, both syntax and word-formation respond to some demand of a languagecommunity, and they are capable of fully meeting the need. In that respect they areabsolutely productive. This also applies to their subsystems. Thus, the systems of Word-Formation Types and Morphological Types are capable of providing a naming unitwhenever a new item, such as Agent noun (or, more explicitly, a noun denoting a personperforming some activity), is required. Then, the clusters of Word-Formation Types andclusters of Morphological Types (see below, Section 2.2.4) ‘guarantee’ the coining of anew naming unit of a specific semantics whenever such demand arises.

All in all, with the advancements in the theory of word-formation in recent decades, theview of low WF productivity has been gradually modified. As a result, I. Plag (1999: 2)

4

could stress that “derivational processes are much more regular than previouslyconceived.”

Our claim that word-formation is absolutely productive because speakers can alwaysprovide a new naming unit when required is not simply a trivial evasion of the problems ofproductivity of word-formation that have engaged morphologists for decades. It is instead arecasting of the issues in terms that should lead to more insight as it brings the analysis ofword-formation closer to the analysis of other linguistic phenomena, particularly syntax.The pragmatic level of analysis has always been implicit in syntax. That speakers cangenerate an infinite number of sentences that they have not previously heard acknowledgesthat speakers do indeed create sentences. There is no effort to keep track of actualsentences or to note all the sentences that could possibly occur but don’t. Syntax is not heldto be any less productive because some sentences do not occur. The communication needsof the speakers is taken as a given before the composition of sentences is analyzed.

Furthermore, the analysis of those sentences does not proceed on the word level; sentencesare not regarded simply as strings of words. Instead sentences are considered to be made upof phrases, which would be another level of abstraction above words. The occurrence ofparticular strings of words, then, has little to do with productive patterns in syntax. Insteadthe issue is whether certain phrase types occur and how they combine together to create asentence. A noun phrase, for example, must be present as the subject of nearly allsentences, yet the composition of that noun phrase – whether a single pronoun, adeterminer and noun, a noun with a complement clause, or something else – isunimportant, so long as the phrase is well-formed. Much less do the individual wordsconstituting the noun phrase matter.

In word-formation studies, however, the pragmatic needs of speakers for new words havelargely been ignored. So too has an abstract level corresponding to the notion of a phrase.Instead, individual formants are considered as productive or not. By focusing on thenaming needs of the speech community and by acknowledging a functional level ofanalysis comparable to phrase structures, a theory of word-formation ought to account forproductivity in word-formation with more coherence.

It is for that reason that this paper proceeds with an onomasiaological approach thataccounts for speakers’ naming needs.

1.2. Potential Words and Naming Needs

The attractiveness of focusing on the naming needs of speakers also comes up with regardto another sticking point of word-formation theory, namely the role of possible or potentialwords. Halle (1973) had introduced the notion of overgeneralization in his generativeaccount of morphology and had called the non-existence of such words in English anaccidental gap. We believe that the notion of ‘accidental gap’ is misleading, and is due tothe purely formal point of view. If this issue is approached from the point of view of thenaming demand of a speech community the non-existence of such words is expedientlyaccountable – they are not needed by the speech community. But what about potentialwords? Should a theory of word-formation account for all words that could be generated orjust those that have? Linguists from Allen to Aronoff to Kiparsky have grappled with thisquestion. The most comprehensive analysis of the relevant problems is given in Bauer(2001) where the relations between actual, existing, established, possible, potential, andprobable words are discussed in detail. Bauer’s ideas may be succinctly summarized as

5

follows:

The notion of existing word raises the fundamental problem of for whom and what such aword exists. We agree with Bauer that an existing word must exist for a speechcommunity keeping in mind a number of problems connected with this approach that arepointed out by Bauer (the lower limit of speech community, non-occurrence of all existingwords in reference works, the identification of the date when a word comes into existence –the first coining or the establishment of the word?, etc.). Bauer (2001: 36) suggests thefollowing definition:

...a word is an existing word from the moment it is first coined...The word may be item-familiar toindividual speakers, without having become part of the norm of the language. A word is establishedonce it becomes part of the norm, that is, once it is item-familiar to a large enough sub-set of thespeech community to make it worth listing in reference works.

The notion of potential word is, in Bauer’s view, closely related with the notion of lexicalgap. Importantly, a coinage only occurs if there is a need, a real or perceived gap in thespeaker’s lexicon: “Productivity is all about potential. A process is productive if it has thepotential to lead to new coinages, or to the extent to which it does lead to new coinages.We are aware of productivity only through the new coinages and the patterns of familiarand unfamiliar words coined by the relevant process” (2001: 41).

Bauer further treats the role of naming with his notion of probable words, which are wordsthat are likely to occur. Bauer suggests that possible word be defined in terms of thelinguistic system while probable word by extra-systemic factors (2001: 42).

The questions of potential words and actual words can be seen to hinge on the role ofextra-linguistic reality in word-formation. Ignoring speakers’ naming needs in favor offormal analysis of the langue gives more importance to potential words. Accounting forthose naming needs gives more importance to actual words. In presenting our approach toproductivity, we will argue in favour of including actual words in productivitycomputations.

The theory we propose for accounting for the naming needs of a community is a cognitiveonomasiological theory. Its fundamental principles are presented in the next section.

2. A cognitive onomasiological theory of productivity

The following approach to word-formation productivity is based on a series of articles anda monograph chapter on this topic, including Štekauer (1994, 1998, 1999, 2001, 2003,2005a, 2005b), and attempts at providing a comprehensive theory of productivity withinthe cognitive onomasiological framework.

2.1. General

It goes without saying that productivity is one of the universal properties of language. It ismost clearly manifested at the level of word-formation because the productivity of Word-Formation Types and Rules and Morphological Types makes it possible to generate a newnaming unit whenever a speech community needs it. From this it follows that word-formation deals with Word-Formation/Morphological Types and Rules which areproductive, that is to say, which, from the synchronic point of view, make it possible to

6

form new naming units whenever need be. Obviously, productivity implies regularity: thisenables language users to understand (in an appropriate context) and use new naming unitsthey have never heard before. In the initial period of existence of a new naming unit,regularity can also be used as a kind of mnemotechnics.

2.2. Main Factors

There are several factors influencing an approach to productivity and the resulting shape ofany theory of productivity:– General theoretical framework. This affects answering the questions like ‘What is the

place of word-formation in the system of linguistic disciplines?’, ‘Is word-formationa part of syntax (the transformationalist hypothesis), a part of morphology (‘classical’structuralist theories), a part of the lexical module (lexicalist approach), or is it anindependent module as proposed, for example, in Štekauer (1998)?’, ‘What is therelation between the individual modules?’

– Scope of word-formation: Do cranberry words, word series like receive, perceive,conceive, pertain, retain, clippings, and acronyms fall within the scope of word-formation? Is compounding the matter of word-formation or syntax?

– Productivity – of what? Productivity of affixes, WF rules, WF processes, the whole WFmodule?

– Attitude to possible, potential, actual and established words, and to the overgeneratingmorphology.

– Are Word-Formation Types/rules productive and regular or is their regularity muchlower than that of syntactic and inflectional rules as assumed by Chomsky (1970)?

– Method of productivity assessment. Should productivity be assessed in abstract terms oris it possible to employ precise mathematical methods? In the latter case, what shouldthe computation be based on? Can we employ absolute numbers or should we relate thecomputation to certain formal or semantic elements? Should the computation be basedon the system level (langue) or speech level (parole)? Or, can these two levels becombined for the sake of productivity computation? Thus, should the calculation bebased on dictionaries or corpora?

2.2.1. Theoretical Framework – the Place of Word-Formation within the System ofLinguistic Disciplines

The cognitive onomasiological theory of word-formation identifies word-formation as anindependent and fully-fledged component as illustrated in Figure 1.

The scheme reflects the relations between the individual linguistic components and withinthe word-formation component itself. It follows from the scheme that the word-formationcomponent is an independent module on a par with any other linguistic module. Thescheme represents the crucial triad of relations: extra-linguistic reality – speech community– word-formation component, thus emphasizing the fact which has been ignored by thevast majority of the mainstream word-formation theories, that is to say, that new words donot come into existence in void (as might follow from purely formal theories). Each act ofnaming responds to a very real and specific naming need (demand) on the part of a member(members) of a particular speech community.

Second, the scheme indicates a direct connection between the word-formation and thelexical components, and an ‘only’ mediated connection between the word-formation andthe syntactic components. This makes this model different from those theories which

7

consider word-formation as a part of Lexicon or a part of Syntax. The relation between theword-formation and the Lexical components is based on close ‘co-operation’. The Lexiconstores all naming units (monemes and complex words, borrowed words, clippings andacronyms) as well as affixes, and feeds the word-formation component with word-formation bases and affixes in accordance with its needs. On the other hand, the word-formation component supplies the Lexicon with new naming units formed in it.

By implication, no new words are generated either in the Lexicon (however, any semanticand/or formal modification of naming units formed in the word-formation component mayonly take place in the Lexicon) or Syntax.3

It should be noted that word-formation concerns the formation of isolated naming unitsrather than their use (which is the matter of syntax). Word-formation treats naming units aslinguistic signs stored in particular semantically and morpho-syntactically definedparadigms in the lexical component. The process of forming new naming units means thatthe new naming units can be subsequently retrieved from the lexicon for the purpose ofsentence formation.

2.2.2. Productivity – O. K., But of What?

Productivity is a term frequently employed by linguists in general, and – like a number ofother linguistic terms – it is quite vague, especially in view of the diversity of its‘applications’. Bauer (2001) demonstrates the ambiguity of this term when he points outthat for some scholars particular affixes (Fleischer) are productive, for others, it ismorphological processes (Anderson) that are productive; for yet others, it is rules (Aronoff,Zwanenburg); for a very few it is words (Saussure); for some it is groups ofprocesses (Al and Booij, Anderson). Bauer (1983) discusses the productivity of acomplete module of grammar; for yet another group of scholars, productivity is a feature ofthe language system as a whole. Bauer (2001) himself prefers to define productivity asa feature of individual morphological processes.

3 For an account of constructions like around-and-do-nothing-ish, leave-it-where-it-is-er lady-in-waiting,pain-in-stomach-gesture, see Štekauer (2001).

Lexical Component

Actual naming units

Affixes(including all relevant

specifications)

Word-formation Component

Semantic level

Onomasiological level

Onomatological level

Phonological levelSyntacticcomponent

8

EXTRA-LINGUISTIC REALITY

SPEECH COMMUNITY

Conceptual level

Figure 1: Word-Formation Component and its relation to other components

Dokulil (1962) also presents several possibilities of examining productivity in word-formation: (i) The productivity of a word-formation formant (affix). Here he distinguishes between

(a) an absolute productivity of a formant, i.e., its applicability in forming new words ingeneral, irrespective of the particular Word-Formation Type it is used in, and

(b) relative productivity of a formant, i.e., its applicability in a specific semanticfunction and/or in a particular Word-Formation Type.

(ii) The productivity of a Word-Formation Type, in which case a WFT functions asa pattern for forming new words.4

(iii) The productivity of a word-formation base.5

4 Dokulil (1962: 72) defines Word-Formation Type as a unity of onomasiological structure (Agentivenouns, bearers of Quality, etc.), lexical-grammatical nature of WF base (deverbatives, desubstantives,deadjectives), and formant (words in -er).

5 Dokulil notes that the productivity of WF base is usually relative: it can mostly be evaluated relative toa particular Word-Formation Type. As such, it is viewed as a condition promoting/reducing theproductivity of a particular WF type (1962: 84).

9

Our approach outlined below discusses productivity at four different, and mutuallycomplementary levels (see Section 2.3).

One of the major disadvantages of various computation methods employed for theevaluation of productivity in word-formation is their limited scope; they are usuallyrestricted to the productivity of affixes. This contradicts the generally accepted scope ofword-formation which also includes other word-formation processes. But even if the focusis laid on affixation the existing methods differ in defining the notion of affix, notably interms of the polysemy – homonymy relation. Both of these facts may significantly distortthe results of productivity computation. The prevailing restriction to affixation processes isalso reflected in the methodology of computing productivity which seems to be tailor-madeto this word-formation process.

It may be proposed (Štekauer 2003) that rather than an affix-driven productivity approacha conception is required which, instead of focussing on items (affixes), ranges over allword-formation processes (WFP) (compounding, prefixation, suffixation, conversion,blending, etc.), i.e., one which overcomes the limitations imposed by affixation inparticular and by the individual word-formation processes in general. What is thereforeneeded is a general WF-Rule-driven theory of productivity covering the whole stock ofcomplex naming units.

The latter approach forces us into the definition of the notion Word-Formation Rule(WFR). Unfortunately, this seems to be another strongly ambiguous term, which, on theone hand, heavily depends on the underlying theoretical background, and, on the otherhand, crucially determines the results of productivity computation. Should a WFR bedefined in Aronoffian (1976) terms as a combination of a base plus affix, or in accordancewith Selkirkean (1982) system based on the maximum level of generalization, such as Xn

Yn Xa for suffixation, Allen’s (1978) Primary Compound Formation Rule: [#X#]N ...[#Y#]N [[#X#][#Y#]], Kiparsky’s (1982) generation of primary compounds by insertionof Y Z into a categorial frame X, i.e., [Y Z]X, or some other formally defined principles?What is the optimum level of generalization in this case? Can a formal definition of WFRsprovide a base for covering all word-formation processes? The major trends in research donot favour a positive answer to this question.

Consequently, since the formal base for productivity computation seems to be unable toprovide a unified footing for all complex naming units, attention should be, in our view,zeroed in on the conceptual-semantic facet. Is such a conceptually and semanticallyoriented theory of WFRs viable? Should it take the form of separation hypothesis proposedwithin the framework of Beard’s (1995) lexeme-morpheme base theory, or is there anyother way of treating WFRs?

2.3. Proposal

The present model departs from a form-based approach to productivity, and proposes toexamine productivity within a particular unifying conceptual category (Agent, Patient,Instrument, Negation, Result of Action, Location, Quality, etc.). This approach followsfrom the onomasiological theory of word-formation: productivity is the matter offormation of new words. Each act of naming (as it follows from the scheme in Figure 1)starts at the conceptual level. It is at this level that the ‘object’ to be named is identified asone falling within the conceptual category of Agent, Patient, Instrument, etc. When the

10

conceptual category of the named object is identified, the naming process proper startswithin which the semantic and morphematic components are identified that will constitutethe “naming structure” of the resulting complex word. Irrespective of the numerousvariations in the ‘naming structure’, all words denoting ‘persons performing(professionally) some activity’ are words denoting Agents. By implication, we may studythe share of the various ‘naming structures’ from different points of view to identify theirrespective Productivity Rates (PRs). In the onomasiological approach, we can identify fourdifferent ‘levels’ of ‘naming structures’, and therefore four levels of productivity:

(i) the productivity at the level of Onomasiological Types (OTs)(ii) the productivity at the level of Word-Formation Types (WFTs)(iii) the productivity at the level of Morphological Types (MTs)(iv) the productivity at the level of Word-Formation Rules (WFRs)

2.3.1. Productivity of Onomasiological Types

The onomasiological model of word-formation (Štekauer 1998, 2001) obliterates thedifferences between the traditional word-formation processes by proposing a unified basisfor the description of word-formation. Such a unified basis makes it possible to objectifythe computation of productivity. This cognitively based model of word-formation, takingthe naming demand of speech-community as its starting point, distinguishes fiveOnomasiological Types ranging over the traditional word-formation processes. They arebased on the criterion of which constituents of the onomasiological (logical-semantic)structure are linguistically expressed at the onomatological (morphematic) level (seeFigure 1). In general, the onomasiological structure includes three basic constituents:

(4) Determining constituent – Determined constituent – Onomasiological baseof the onomasiological of the onomasiologicalmark mark

where Onomasiological base corresponds to the head of a complex word, and thedetermined constituent of the Onomasiological mark generally stands for the concept ofACTION. Then, the individual Onomasiological Types can be exemplified as follows:

In Onomasiological Type 1, all three onomasiological structure constituents, i.e., the base,the determining and the determined constituents of the mark, are linguistically expressed atthe onomatological level by being assigned morphemes with the corresponding meaning.This operation is labeled as the Meaning-to-Seme-Assignment principle (MSAP):

(5) truck-driver (A Person (Agent) operates (Action) a vehicle (Object)Object – Action – Agenttruck drive er

(6) house-keeping (The Process of performing some Action aimed at anObject)):

Object –Action – Process house keep ing

(7) signal-generator (Instrument for an Action producing some Result)Result – Action – InstrumentSignal generate or

11

In Onomasiological Type 2, the determining constituent of the onomasiological mark isleft unexpressed:

(8) Factitive – Action – Agent 0 write er

(9) Object – Action – Instrument0 spinning wheel

In Onomasiological Type 3, the determined constituent of onomasiological mark is leftunexpressed:

(10) Result – Action – Agent novel 0 ist

(11) Patient – State –Evaluation (Diminutive) dog 0 ie

(12) Temporal Stative – State – Patientsummer 0 house

In Onomasiological Type 4, the onomasiological mark is simple and unstructured, i.e., itcannot be divided into the determining and the determined constituents.

(13) Negation – Qualityun happy

(14) Quality – Stateblue-eye ed

(15) Repetition Action re gain

Onomasiological Type 5 (onomasiological recategorization) concerns conversion, and themethod of representation of semantic relations between the members of conversion pairs isillustrated by the following examples:

(16) bondN – bondV: SUBSTANCEResultACTION(in the meaning of a joint)Interpretation: Substance as a Result of Action

(17) switchN – switchV: SUBSTANCEInstrument/ResultACTION (in the meaning of a device for completing or breaking an electric circuit)Interpretation: Substance as an Instrument of Action

(18) insertV – insertN: ACTIONObjectSUBSTANCEInterpretation: Substance as an Object of Action

12

(19) timeN – timeV: CIRCUMSTANCETemporalACTIONInterpretation: Action in terms of Temporal dimension

(20) clearA – clearV: QUALITYResultACTIONInterpretation: Action Resulting in a certain Quality

As indicated above, the present model distinguishes five Onomasiological Types rangingover all productive methods of forming new complex words. Since they are based on thecriterion of which constituents of the onomasiological structure are linguistically expressedat the onomatological level, the determination of their respective productivities is animportant indicator of the preferences of language users (or better, coiners) in terms ofemploying different cognitive processes underlying the act of naming, on the one hand,and the different ways of their linguistic representation, on the other. The productivitycalculation at this level may indicate which of the two universal, contradictory tendencies,i.e., economy of expression and explicitness of expression (comprehensibility),dominates in a particular language (area). Here we face two gradual oppositions:

(i) Onomasiological level (a) Onomasiological Types 1–3 (complex onomasiological structure) (b) Onomasiological Type 4 (simplified onomasiological structure) (c) Onomasiological Type 5 (absence of onomasiological structure)

(ii) Onomatological level(a) Onomasiological Type 1 (complex morphematic representation of complex

onomasiological structure) (b) Onomasiological Types 2 and 3 (economized morphematic representation of a

complex onomasiological structure(c) Onomasiological Type 4 (economy due to onomasiological structure)(d) Onomasiological Type 5 (absolute economy – no morphematic representation).

As indicated above, productivity of the individual Onomasiological Types is given by theirrespective share of all the complex words that belong to a particular conceptual category(e.g., Agent). From this it follows that the Onomasiological Type Cluster is 100%productive with regard to a particular conceptual category as it can ‘produce’ a wordbelonging to that particular conceptual category whenever a (member of a) speechcommunity needs to give a name to an object belonging to this category.

2.3.2. Productivity of Word-Formation Types

A more specific level is represented by WF Types. The computation of productivity of WFTypes is also related to a particular conceptual category. This makes it possible to includein the computation of the productivity of, for example, Agent names complex words ofdifferent onomasiological structures, hence different WF Types (for example, Object –Action – Agent (woodcutter); Action – Agent (writer); Location – Action – Agent (street-fighter); Factitive – Action – Agent (novel writer); Instrument – Action – Agent (anthrax-killer); Manner – Action – Agent (slam-dunker); and a number of other possible WFtypes).

All of these WF Types may be used to coin new complex words falling within one and thesame conceptual category (Agent, in our example), and therefore represent a single Word-Formation Type Cluster (WFTC). Any WFTC is – with regard to the particular

13

conceptual category – 100% productive. Therefore, the productivity of the individual WFTypes may be computed internally, within the WFTC, as a share of the individual WFTypes of the total number of complex words belonging to the given WFTC.

2.3.3. Productivity of Morphological Types

Any WF Type may have various morphological representations (wood-cutter (=N+V+er)– novelist (N+ist) – writer (V+er) – cheat (conversion) – oarsman (N+s+man) –transformational grammarian (A+N+ian) – bodyguard (N+N), etc.). All of these differentmorphological structures represent various Morphological Types. Since they are used tocoin new complex words falling within one and the same conceptual category (Agent, inour example), they represent a single Morphological Type Cluster (MTC). Any MTC is –with regard to the particular conceptual category – 100% productive, and the productivityof the individual Morphological Types may be computed internally, within the particularMTC.

2.3.4. Productivity of Word-Formation Rules

Word-Formation Rules are constituted by the unity of WF Types and MorphologicalTypes. Thus, the conceptual category of Agent category may be exemplified, inter alia, bythe following WF Rules:

(21) a. Action – Agent Verb er (driver)

b. Instrument – AgentNoun (s) man (oarsman)

c. Object – Action – AgentNoun Verb er (wood-cutter)

From this it follows that the WFR is constituted by the unity of the onomasiological andonomatological structures.

2.3.5. Justification

The reason for preferring this approach to the calculation of Productivity Rate is that (a) it makes it possible to examine productivity from different viewpoints reflecting both

linguistic and supralinguistic levels;(b) it takes into consideration all new words (not just some WF processes like affixation);(c) it restricts itself to actual words (i.e. words coined in response to the needs of a

particular speech community) in order to avoid the one-sided formalism of themainstream discussion on word-formation.

From the previous discussion it follows that productivity is conceived as an implementedcapacity reflecting the naming needs of a particular speech community. As suggestedin Štekauer (1998, 2001), what seems to be crucial is that by coining a new word inresponse to the specific demand of a speech community the particular language manifestsits productive capacity to provide a new, well-formed linguistic sign by employing itsproductive types/rules whenever need arises. By implication, inclusion in the model of the

14

extra-linguistic factor (speech community) makes it possible to eliminate the notion ofovergeneration.6

This approach is in accordance with Bauer who maintains that “[t]he fact remains ... thatthe production of new words may be the only evidence the observer has of this potential,and the lack of new words appears to deny the potential” (2001: 21) and that “...words areonly formed as and when there is a need for them, and such a need cannot be reduced toformal terms” (2001: 143). In principle, the conception of productivity as implementedcapacity corresponds with Bauer’s (2001) notion of ‘profitability’.

Obviously, the proposed model of computing the productivity takes dictionaries as itsbasic source of data. This is not viewed as its drawback. It is believed that the method canbe advantageously applied to the determination of productivity in selected lexical fields(sciences, sports, culture, etc.) as captured – generally fairly well – in a number of special-purpose dictionaries. It can also be applied to identify the latest trends in coining newnaming units thanks to the dictionaries of neologisms and/or lists of new words aspublished, for example, in American Speech. Since productivity changes are not the matterof weeks, months, nay even one or two years, the time lag of covering these trends bydictionaries does not seem to be a relevant objection against this method. Moreover, it maybe proposed that studying the general productivity should be subordinated to thedetermination of the productivity in the individual spheres of life as captured by special-purpose dictionaries. Namely, it may be postulated that the situation and the trends incoining new words in the fields like, for example, medical research versus fashion pursuedifferent trajectories. From this it follows that any generalizations based on unequalamount, structure, and range of data may be fairly misleading.

The model proposed can be illustrated by the results of a case study focussed on the namesof INSTRUMENTS (including tools, devices, machines, equipment, appliances, implements,apparatus, etc.) in the English-Slovak Technical Dictionary by A. Caforio (1996) under thearbitrarily selected letter “S”. The analysis of 192 naming units indicates that – out of thefive Onomasiological Types – the most productive is Onomasiological Type 3, with over55% Productivity Rate, followed by Onomasiological Type 1 with 28% PR,Onomasiological Type 2 with 12.5% PR, and conversion (almost 5% PR). From this itfollows that there is a very strong tendency to morphematic representation of the Actionalsemantic component of the onomasiological structure (over 80% of all naming units).

At the level of WF Types, the most productive is the [ActionPurposeInstrument] type withmore than 55% PR, followed by [Object–ActionPurposeInstrument] with 15% PR. Thelimited sample indicates the tendency for Instrumental names to leave the determiningconstituent of the OM unexpressed, thus producing less specialized terms to the benefit ofhigher-level generalizations, and – by implication – broader applicability of theinstrumental naming units.

At the level of Morphological Types, the [ActionPurposeInstrument] type, for example, isdominated by the [stem + -er/-or] MT (e.g. sensor, slipper, selector) the productivity ofwhich amounts to almost 72 %. The remainder is represented by the [stem – stem] MTwith over 25 % productivity (e.g. suction funnel, search coil, summation instrument), andconversion (e.g. slide, rule). Again, important conclusions can also be drawn at this lowest6 Which means that ‘our’ word-formation component (unlike, for example Halle’s (1973) and Allen’s

(1978) does not ‘generate’ possible, but ‘non-existing’ words, i.e., it does no more than is actually neededby a speech community.

15

productivity level, i.e., the most frequently employed Morphological Type for Instrumentalnames is one with the -er/-or suffix.

This outline indicates that the proposed model makes it possible to draw relevantconclusions by interrelating all word-formation processes at various levels ofgeneralization, depending on the specific needs of analysis.

2.3.6. Word-Formation as Creativity within Productivity Constraints

The terms ‘creativity’ and ‘productivity’ are usually understood as mutually excludingprinciples in coining new words. While productivity is said to be rule-governed, creativityis conceived of as any deviation from the productive rules. In the present context, creativityis used in a different meaning in which it is complementary with productivity. First, thelogical spectrum (conceptual level) does not necessarily lead to one singleOnomasiological Structure. For illustration, if we try to form a naming unit for ‘a personwho meets space aliens on behalf of the human race’ the logical spectrum may yieldvarious word formation types, such as Theme – Action – Agent, Location/Theme– Action– Agent, Location – Action – Agent, Object/Location – Action – Agent, Object – Action –Agent. Second, these different Word-Formation Types may be assigned variousmorphological realizations by the MSAP principle, for example,

(22) a. Theme – Action – Agenthuman race representative (Onomasiological Type1)homosapience representative (Onomasiological Type 1)

b. Location/Theme – Action – Agentearth-representative (Onomasiological Type 1)earth ambassador (Onomasiological Type 2)world ambassador (Onomasiological Type 2)

c. Location – Action – Agentintergalactic diplomat (Onomasiological Type 2)interstellar diplomat (Onomasiological Type 2)

e. Object/Location – Action – Agentextra-terrestrial greeter (Onomasiological Type 1)space alien meeter (Onomasiological Type 1)outerspace wellcomist (Onomasiological Type 1)

f. Object – Action – Agentcontactee (Onomasiological Type 3)greeter (Onomasiological Type 3)7

Example (22) thus illustrates what can be labeled as creativity within productivityconstraints. It illustrates, on the one hand, different onomasiological realizations ofa particular logical spectrum, and, on the other hand, different onomatological realizationsof various onomasiological structures. It is the interaction between the conceptual,onomasiological, and onomatological levels which – within the limits of productive typesand rules and the relevant constraints – provides certain space for a creative approach toword-formation (as it follows from several options in our example). This meaning ofcreativity emerges from a cognitive onomasiological approach. The inclusion of speechcommunity in the model and viewing each new naming unit as a result of a very specificand real act of naming by a coiner makes it possible to reflect in the present model

7 The examples in (22) were proposed by Native speakers.

16

individual preferences, the influence of one’s age, education, and profession, as well asone’s linguistic family background (in a bilingual setting), fashionable trends, etc., i.e., thesociolinguistic factors which may affect the application of the MSAP in those cases thatprovide more than one option. Thus, it is in this sense of ‘creativity within productivityconstraints’ that the presented onomasiological approach treats word-formation, and inparticular, the relation between productivity and creativity. This brings us to anexperimental research aimed at the application of the ‘multilevel’ computation ofproductivity and at demonstrating the validity of the concept of word-formation ascreativity within productivity constraints. For that purpose, we will presentsociolinguistically oriented evidence.

3. Sociolinguistic Research into WF Productivity

3.1. General

It is generally accepted that word-formation processes are never totally unrestricted, andeven the most productive affixes seem to be subject to certain structural constraints (Plag1999: 35). In the literature on word-formation, a number of restrictions upon productivitywere mentioned. In addition to the ‘traditionally’ adduced systematic constraints,8

including phonological, morphological, lexical, syntactic, and semantic ones, both Bauer(2001) and Plag (1999) list some other, extra-linguistic factors, including

(a) Pragmatics (because of denotation and connotation of some WF patterns they are notin common use, e.g. suffix -some in words like twosome. In principle, twenty-five-some may be possible, but it is not usual because we do not usually operate withgroups of 25 people);9

(b) Aesthetics, e.g., word-length;(c) ‘Accidents of cultural history’, e.g., a person whose job is to sell things happens not

to be called seller since salesman/saleswoman is the established form. In these casesthere is no linguistic reason for the current usage, it just so happens that a particularpossible form has not become part of the norm;10

(d) Failure of hypostatisation: Coining a new word presupposes that there is an entity tobe denoted by the new word. If there is no such entity, there is no need for a word.11

It appears, however, that in spite of abundant literature on productivity constraints, there isat least one factor that has been neglected and that deserves attention of morphologists, inparticular, the sociolinguistic factor. We believe that productivity of OnomasiologicalTypes, Word-Formation Rules/Types, and Morphological Types is also affected bysociolinguistic factors which may be divided into two groups:

8 For a comprehensive review of various restrictions as well as blocking theories see Plag (1999), Bauer(2001), and Rainer (2005).

9 For a review of pragmatic factors (fashionability, demand, attitudinal function, hypostatization,nameability) see Plag (1999).

10 This is not to say that constraints on productivity of any type are absolute. In the case of ‘accidents ofcultural history’, for example, the blocking principle can be ‘overpowered’ by a particular Word-Formation Type gaining in productivity (for any reasons, including, inter alia, those concerning voguishuse).

11 That the situation can change can be illustrated by *loather. Bolinger (1975: 109) notes that this word isnot an actual word of English not because it cannot be formed, but because “we have no use for it. Whatretinue of people would it designate?”. Bolinger’s view is also referred to by L. Bauer (2001: 43).However, Ingo Plag, (personal communication) drew our attention to “numerous nice attestations of thisword on the internet (two even in dictionaries).”

17

(i) Horizontal factors, including the previous linguistic experience. This factor plays itsrole in multinational countries, such as the USA, Australia, Great Britain, and in fact,a number of other countries due to the growing migration. There are millions of peoplewhose grandparents, parents, or they themselves were born and have lived in alinguistically different environment. Interestingly, while the factor of linguistic interference has been a topic of manytreatises focussed on grammar, pronunciation, etc., the issues of interference in word-formation has not been – to our knowledge – studied yet.

(ii) Vertical factors, including various social strata, education levels, professions, etc. Itgoes without saying that these factors affect the extent of actively and passivelymastered vocabulary of a speaker, and hence influence his/her linguistic behaviour,which cannot – in our view - remain without effects upon the formation of newnaming units.

Given these postulates, it may be proposed that any act of word-formation is a kind ofintersection of three factors:

(i) the pressure of the productivity of individual Onomasiological/Word-Formation/Morphological Types and Word-Formation Rules within the respectiveconceptual-semantic clusters;

(ii) the extent of experience (including no experience) with a native language other thanEnglish;

(iii) vertical sociolinguistic factors.

As it follows from experimental data, the latter two factors have their say at theonomasiological and the onomatological levels of the word-formation model (Figure 1),that is, at the level of conceptually identified logical-semantic structure establishing thebasis for the act of naming, and at the level of its linguistic expression (assignment of WFbases and affixes to semes). It is these two levels that provide – as we believe – sufficientspace for the operation of extralinguistic factors. In other words, it is at these levels thatone’s naming preferences may be implemented as the above-mentioned sociolinguisticfactors may affect a coiner’s selection (influenced by his/her former mother language word-formation patterns, education, extent of his active vocabulary, the register used in his/hersocial stratum and occupation, etc.) of one or the other affixation type, a verbal compoundtype, a non-verbal (primary) compound type, a conversion, blending (to use traditionalterminology), etc.

3.2. Experimental research

3.2.1. General

Our experimental research was aimed at identifying the validity of our hypothesisconcerning the influence of sociolinguistic factors upon the productivity in word-formation, in particular, the role played by linguistic background, education, andprofession.

For the sake of our experiment, a questionnaire was developed (see Appendix 1). The basictask of the informants was to give names to ‘objects’ for which there did not exist anycorresponding names in English at the time of our experiment. To avoid inconsistency, all

18

the objects to be named were conceived of as Agents. Our decision to concentrate onAgents was motivated by a relatively large number of different rules that make it possibleto coin Agent names.

To avoid any distortion of results due to one-sided/inappropriate formulation of theexperimental task, we decided to provide our informants with a questionnaire consisting oftwo basic parts, with the first part including three differently formulated naming tasks. Thefirst was a selection task. Each object to be named was briefly characterized, e.g., ‘a personwho frequently interrupts other people when they are talking’. The characterization of theobject of naming was followed by a set of options. In this particular case, they includedinterrupter, interruptist, butt-in, butter-inner, cutter-in, cutman, interposer, and a fewothers. In addition, the final option in each set was a blank line which could be filled in ifan informant did not find any of the options offered to be a suitable way of naming theobject.

Task 2 differed from Task 1 in not containing any options. The informants had to proposetheir own naming units based on a brief specification of the object to be named, forinstance, ‘Suppose that space aliens were about to land on Earth for the first time. Whatwould you call a person who was supposed to meet them as a representative of the humanrace?’

Task 3 replaced wording by a drawing of a situation in which an object performs someunusual activity, for example:

Figure 2

Part 2 was of a different nature. The purpose was to identify any possible differences inlinguistic (naming) behaviour of various groups of language users with respect tounproductively coined naming units, i.e., naming units which were formed in defiance ofrelevant productivity constraints. For this purpose, five naming units were formed:engroupment, thinnen, swimmee, sleepable, and satisfactority. These naming units violatedifferent productivity constraints.

The views of the suffix -ment, by means of which engroupment was formed, differ. WhileBauer (1983: 49) maintains that this suffix does not seem to be productive any more, otherslike Plag (1999: 72-75) and Adams (2001: 28) demonstrate that -ment is low productive. Itfollows from Plag’s analysis that the best candidates for -ment derivation are verbs endingin the suffix -en, having a disyllabic bases with stress on the second syllable, and with a

19

prefix like be-, en-. As Plag notes, “[t]he preference for prefixed stems is especiallyobvious with stems containing the prefixes eN- and be-, which seem to take -mentobligatorily” (1999: 73). Thus, engroupment is possible as it meets the specifiedrestrictions. On the other hand, the specific WF pattern is low productive. In addition,engroupment violates the ‘Avoid Synonymy Principle’ (Kiparsky 1982) because its placein the system has already been filled in with group and grouping.

For thinnen the constraint is different: the inchoative suffix -en only attaches tomonosyllabic stems if and only if they end in an obstruent, optionally preceded bya sonorant (Halle 1973). Furthermore, the suffix -en does not seem to be productive anymore (Lieber 2004: 76).

The selection of swimmee based on the [V+ -ee]N pattern requires a more detailedexplanation. As suggested by Barker (1998: 708), the suffix -ee can be viewed as acounterpart of -er, and “it is possible to entertain the hypothesis that the conditions for useof -ee are defined negatively, in contrast to those for -er: -er picks out subject participants,and -ee covers everything else.” As he, however, notes this hypothesis faces the problem ofthe existence of a considerable number of -ee nouns referring to subject participants. Thisis confirmed by Lieber (2004, 2005), who points out that while -er nouns “most often formpersonal agent nouns, and -ee most often forms patient/theme nouns, not infrequently wefind precisely the opposite situation, where -er and its cohort form patient nouns and -eeagent or at least subject-oriented nouns” (2005: 404).

Moreover there are instances of both -ee and -er attached to the same WF base havingsynonymous meaning (escapee/escaper, absentee/absenter, arrivee/arriver, etc.) (Barker1998: 709). Based on the analysis of a large corpus, Barker arrives at a conclusion thatthere are at least three types of -ee derivations that are productive: direct object, indirectobject, and subject. Our swimmee is the subject type. This possible naming unit safelymeets two of three of Barker’s (1998) semantic conditions imposed on productive -eederivation. First, it meets the condition of sentience12 of the referent referred to byswimmee. Second, it meets the semantic constraint of ‘episodic linking’, according towhich “the referent of a noun phrase headed by an -ee noun must have participated in anevent of the type corresponding to the stem verb” (1998: 711). In this particular case,swimmee participates in a swimming event. Problematic is the third semantic constraint,defined as “a lack of volitional control on the part of its referent either over the occurrenceor the duration of the qualifying event itself or (given a punctual qualifying event) over itsimmediate direct consequences” (1998: 717). Being a subject type swimmee refers toAgent, but the category of Agency implies volition. Thus, for a swimmee to preserve the‘lack-of-volition’ constraint it would have to mean the action into which a swimmingperson is forced somehow – in contrast to swimmer, who, in principle, does his/her activityvoluntarily, fully based on his will.13

It follows from the above discussion that swimmee is a possible naming unit that can beproduced by a productive WF rule. What made us include this word in the ‘unproductivity’test is the much more productive competitor, the -er-based pattern that underlies theexisting and well-established Agent noun swimmer. Thus, while the -ee Agent noun ispossible (other meanings of -ee nouns, such as Patient or Theme can hardly be expected aswas also manifested by our subsequent experimental research) it is blocked on a general12 Which means reference to an Animate entity.13 Even this is not quite so: a professional swimmer training is a hard drill under the control of a coach, and

not always in accordance with the will of the swimmer.

20

Agentive level by a much more productive WF rule that has already produced a firmlyestablished (institutionalized) naming unit swimmer. On a fine-grained semantic level, theblocking is eliminated by the ‘volition – lack of volition’ opposition. The question behindthe inclusion in the experimental ‘unproductivity’ research of swimmer was whether theinformants (native speakers) will perceive this kind of semantic distinction.

As it will follow from the results of our experiment, while almost each of our nativespeaker informants was able to propose a sentence in which they used swimmee, the vastmajority of them find swimmee to be ‘extremely unlikely’ or ‘somewhat unlikely’ justbecause the uses proposed were in the absolute majority of cases connected with the Agent-based interpretation that did not distinguish the ‘volitional’ constraint. In fact, none of ournative speaker informants referred to the volitional aspect of the swimming action. As suchthe meaning of swimmee was to the vast majority of the informants blocked by swimmer.14

The suffix -able, which occurs in our naming unit sleepable, does not meet the traditionallyadduced restrictions, summarised in Anderson (1992: 186):

(23) WFR: [X]V → [Xәbl]Adj

Condition: [X]V is transitive (i.e., [+_NP])Syntax: ‘Object’ argument of [X]V corresponds to ‘Subject’ of [Xәbl]Adj

Semantics: ‘(VERB)’ → ‘capable of being VERBed’

Sleepable is intransitive and there is hardly any acceptable reading that would meet thesyntactic condition. From the semantic point of view, it rather features a ’propertymeaning’, to use Plag’s (2004) term. Importantly, as noted by Plag (ibid), “the formsexhibiting the property meaning are in a clear minority. In fact, this pattern has ceased to beproductive as early as the 17th century…” All these facts imply that sleepable is a goodcandidate for our unproductivity test.

Finally, satisfactority violates the constraint according to which -ity is only productive (ispotentiated – Williams 1981) in combination with the productive -able function, whichmaps transitive verbs to adjectives. By implication, the domain of the -ity function fity is thefunction fable, and its range is the composed function fability (Raffelsiefen 1992).

This task was thus aimed at recognizing the ‘sensitivity’ of different groups of speakers toproductivity constraints and the ‘inappropriately’ coined words. The informants were givena five-degree scale, including the options of ‘extremely unlikely’, ‘somewhat unlikely’,‘likely’, ‘very likely’, and ‘extremely likely’. In addition, they were asked to give anexample of a sentence, including an ‘unproductively’ coined word.

The questionnaires were collected (and the informants were approached) in various ways,in particular through personal contacts, through our students and friends in English-speaking countries, through the Internet LinguistList service, and finally, through a special-purpose www.page. It follows that it was fairly difficult to meet the initial goal of havingthe individual subgroups per profession, occupation, and different linguistic backgroundevenly distributed. In any case, we believe that the extent of the sample made it possible to

14 Instead of the volition-related constraint, proposed by Barker, some of the informants distinguishedbetween swimmee and swimmer in terms of ‘swimming skill’, mostly in favour of swimmer. The unequalskill, ability, and capacity per se (in any activity) do not, however, seem to be a sufficient justification fora productive WF process. This would lead to an extremely high number of naming units and aconsiderable overload of a language user’s memory.

21

accomplish the basic objectives of our experimental research and to draw relevantconclusions. The sample of informants was divided into two groups, the native speakerswhose parents were born in an English-speaking country (language proficiency A in thequestionnaire), i.e., those who were not influenced at home by immediate contact with adifferent language; and native speakers whose parents were not born in an English-speaking country plus non-native speakers living in an English-speaking country (languageproficiency B and less in the questionnaire), i.e., those whose English competence had tocope with the influence of another language. The former group (speakers unconnected withanother native language) has been subdivided accordingly into various subgroups byoccupation (students, educators, ‘other’ professions), and by education (high school,college, graduate). The latter group (speakers connected with another language) has beendivided into groups based on the morphology of noun, namely, synthetic/agglutinative,synthetic/fusional, analytic/isolating, and polysynthetic. Due to very low numbers ofinformants (three), the polysynthetic group was not taken into consideration.

3.2.2. General Analysis

3.2.2.1. Native Speakers

The experiment encompassed 145 native speakers from various English-speakingcountries, mostly from the USA. The total number of ‘responses’ amounts to 4 tasks times5 subtasks per each, which gives 20 responses per informant, which, ideally, adds up to2,900 ‘responses’ in total. However, the actual number of responses is smaller (1,531) fortwo major reasons:

1. Not all of the informants completed all sub-tasks.2. Some informants did not specify all relevant data within the demographic informationsection of the questionnaire, and therefore their replies could not be taken into account inall parts of our analysis;

The number of 1531 responses was further reduced down to 1300 relevant responses thatbecame an object of our analysis. The difference of 231 responses that were eventuallyeliminated from consideration follows from the fact that our research was focused on theproductivity in word-formation, and therefore all those naming units proposed by theinformants which resulted from sources other than productive word-formation wereeliminated from the scope of analysis:

(a) They were mostly proposals based on semantic shift of an already existing word –in which case no new naming unit comes into existence.

(b) In addition, only those naming units were taken into account that indicated themeaning specified by the descriptive wording or drawing. Therefore, wedisregarded proposals like ambassador, welcomer, ET, and President of the USAfor ‘a person meeting space visitors’; risk-taker, show-off, crazy, mad, retard,daredevil, weirdie for ‘a person riding on car-body top’; comedian, platinumrecord, idiot, cruel, and joker for ‘a person frequently joking about blondes’, weird,to denote ‘a person who dials a telephone number with a feather’, zoologist for‘someone who does research about spider webs’; perfectionist, fussy, meticulousgardener, biologist, and frowny face for ‘a person cutting grass with a knife’; show-off, Michael Jordan and monkey for ‘a basketball player who always hangs onto therim after a slam-dunk’, macho and acrobat for ‘a person lifting weights on a crane’;time-killer for ‘a person tying shoelaces to customers’, etc. They are either ‘mere’

22

extensions of the original meanings, or are too general to say anything relevantabout the actual mission/activity of the individual objects to be named.

(c) We also eliminated those naming units that do not correspond with the productiveWF types in English or are ungrammatical in any other way, for example,arachologue, slam duckle, pedlacier, person flying over car, believer in miracles,researcher on spider webs, etc.

(d) Finally, the following analysis does not take into consideration the names ofPatients15 that also occurred in the experiment. This leaves us with 1,300 responses.

Importantly, since the focus of our research is on WF productivity rather than on individualnaming units the following analysis concentrates on types and rules. Brief comments onsome interesting cases of individual naming units are given in 3.2.6.

3.2.2.2. Productivity of Onomasiological Types

As already suggested above, there are two contradictory tendencies in language, thetendency to the economy of expression and the tendency to the explicitness ofexpression (clarity of communication). If we analyse the results in view of scale (i)specified in Section 2.2.4.1, that is to say, in view of the complexity of onomasiologicalstructure, we find out that the total number of responses for the onomasiologically‘explicit’ types 1, 2 and 3 is 1,272 and that for the ‘non-explicit’ type 5 is 28 (Table 1). Nonaming units were based on Type 4. From the point of view of scale (ii), in particular, theexplicitness of the onomatological level, it may be concluded that the number of explicittypes (Type 1) roughly corresponds with the number of ‘economic’ types (51.54% :48.46%).

Moreover, the central role is played by those Onomasiological Types (Types 1 and 2)whose determined constituent (i.e., the Action-representing constituent) is explicitlyrepresented by a morpheme. In total, they represent 75% of all naming units. This result isnot surprising because it is this constituent that is vital to the understanding of new namingunits. The Actional constituent namely relates the onomasiological base with thedetermining constituent of the mark in Onomasiological Type 1 thus significantlycontributing to the interpretability of such naming units. Also in type 2, the determinedconstituent clearly indicates the ‘Action’ of the Agent represented by the onomasiologicalbase. Thus, for example, the determined constituent of the mark (surf) in roof-surfer clearlyand unambiguously identifies the relation between the polar members of theonomasiological structure, i.e., roof and -er, and makes the interpretation of this namingunit easy. On the other hand, a naming unit falling within the scope of type 3, i.e., roofer,makes the process of meaning interpretation pretty demanding just because there is noActional constituent that would appropriately identify the relation between the polarmembers of the onomasiological structure represented by the morphemes roof and -er. Asa result, the number of possible interpretations of roofer is considerably high.16

The Onomasiological Type 2 variant of the same ‘object’, i.e., surfer is more valuable interms of meaning predictability than the type 3 variant because it identifies the actualAction of the Agent.

15 Patient is here defined as ‘Bearer of State’. Examples from our experiment include car-topped guy,obsessionist, clone, etc.

16 For a theory of meaning predictability of naming units coming into existence by word-formation processessee Štekauer (2005a).

23

There is no Agentive naming unit of Type 4. In general, however, this OnomasiologicalType is highly valuable in terms of easy interpretation because the direct connectionbetween the unstructured mark and the base at the onomasiological level gives no chancesfor multiple interpretations. This can be exemplified by a Patient name that occurred in ourresearch, sub-clone, where the mark sub- directly specifies the Quality of Patient clone.

Given our results, however, type 4 does not seem to be a productive type for Agent namesfor the simple reason that it usually specifies the Quality rather the Action performed byAgent.

No. of responses PR (%)OT1 670 51.54OT2 299 23.00OT3 303 23.31OT4 0 0.00OT5 28 2.15Total 1300

Table 1: Predictability Rate of Onomasiological Types (native speakers)

3.2.2.3. Productivity of Word-Formation Types

Since the experiment examined the naming preferences of English speakers in the fieldAgents (1300 responses), i.e., persons performing some Action, the dominant position ofWFTs [Object–Action–Agent], [Action–Agent] and [Theme–Action–Agent] is notsurprising: since Agents are human beings performing some Action, the presence of thedetermined constituent of the onomasiological mark (which, as we already know, standsfor Action in general) is expected.

The most productive types in our research indicate two basic tendencies in the naming‘behaviour’ of native language users. First, they select a more general naming unit becausethey either wish to increase its extension, for example, to avoid the exclusion of someunpredictable special-purpose cases, or, because the scope of Agent’s Action is vaguelydefined. In our experiment, one such reason which contributes to the productivity of the[Action–Agent] WFT to the detriment of a more explicit type was the fact that some of ourinformants were not quite sure about the specific nature of the broadly conceived activity tobe named – for example, the ‘grass-cutting’ and the ‘shoe-lacing’ drawings (cf. Appendix1).

Second, in the majority of cases, there is an effort of native language users to be morespecific (circumstances-permitting) and express those categories which are inherentlyrelated to Action, such as Object of Action, Instrument of Action, Theme of Action,Location of Action, Time of Action, etc. Consequently, there is no wonder that the[Object–Action–Agent] WFT is the most productive in our sample, with 416 responses,yielding the Productivity Rate of 32.0%. The PR of the [Theme–Action–Agent] WFT is17.3%, [Instrument–Action–Agent] 10.4%, and [Location–Action–Agent] 9.1%. ThePR of the above commented, more general WFT [Action–Agent] is 23.4%. These fiveWFTs represent about 92% of all Agentive naming units in the native-speaker group ofinformants, which clearly indicates their high productivity, on the one hand, and a minorrole played by the remaining WFTs.

24

PR (%)Object–Action–Agent 32.00Action–Agent 23.38Theme–Action–Agent 17.31Instrument–Action–Agent 10.38Location–Action–Agent 9.08Table 2: Predictability Rate of Word-Formation Types (native speakers)

3.2.2.4. Productivity of Morphological Types17

The number of options for Morphological Types is not large as it is limited by thecombinability of stems and affixes, subdivided (in English) into prefixes and suffixes.Since Agentive functions are primarily expressed by suffixes in English, thoseMorphological Types are more productive which combine stems (S) with suffixes.18 Whilethe most productive Morphological Type [S+S+suffix] (PR=47%) corresponds with theexpectations stipulated in 3.2.2.1.2 above, i.e., that a more explicit structure is preferred,the distribution of MTs in terms of implicit and explicit structures is roughly balanced,with the two most productive MTs being two-constituent structures [S+suffix] (37%) and[S+S] (11%), respectively. On the other hand, the MTs with a suffix in the role ofonomasiological base clearly prevail with 85% PR.

(PR%)S + S + suffix 46.67S + suffix 37.41S + S 11.29Table 3: Predictability Rate of Morphological Types (native speakers)

3.2.2.5. Productivity of Word-Formation Rules

Word-Formation Rules result from the operation of the Morpheme-to-Seme-AssignmentPrinciple, which means that they reflect the interrelation between the onomasiological andthe onomatological levels. As such, they should reflect the basic tendencies in the domainof Word-Formation Types (onomasiological level) and Morphological Types(onomatological level). Therefore, since the most productive WFT is [Object–Action–Agent], since the most productive MT is [S+S+suffix], since Objects of Action are usuallyexpressed by nouns, and, finally, since it is generally known that the -er suffix is the mostproductive Agentive suffix in English (much more productive than its competitors, like-ist, -ant, -ee, -ian) one may expect the dominating position of the following WFR

(24) Object–Action–AgentN V -er

And actually, the results bear out this postulate, as it follows from Table 4. Theprominent position of the ‘-er-for-Agent’ structures is strengthened by the fact that the fourmost productive WFRs are of this sort, with their share of the total number of Agentive

17 The cases of OT5 are not included for obvious reasons. Thus, the number of responses taken intoconsideration for Agents is 1 275, those for Patients is 165.

18 The naming units with man in the position of onomasiological base are classified as stem-based units inspite of the fact that a number of authors treat this element as semiaffix.

25

WFRs exceeding 50%. In addition, it can be concluded from Table 4 that WFRs with stemsin the function of an onomasiological base are far from being high-productive in English.

PR (%)Object–Action–Agent (N V -er) 20.92Action–Agent (V -er) 17.85Theme–Action–Agent (N V -er) 10.62Location–Action–Agent (N V –er) 7.08Object–Action–Agent (N 0 –ist) 4.38Instrument–Action–Agent (N 0 -ist) 4.23Action–Agent (V N) 2.69Instrument–Action–Agent (N V -er) 2.23

Table 4: Productivity Rate of Word-Formation Rules (native speakers) – Agents

3.2.3. Influence of Occupation

3.2.3.1. Analysis of the Experimental Data

Taking the general picture, discussed in Section 3.2.2, as a reference point, we can proceedto the comparison of the data obtained for the individual groups of informants, based ontheir occupation. The available sample of informants necessitated their division into threegroups, in particular, students, teachers, and ‘other’ professions. The sample included 60students, 35 educators, and 50 ‘other’ professions who produced 1531 (1300 for Agentsand 231 for Patients) responses in total. The latter group of occupations was originallysubdivided into those of civil servants, natural scientists and engineers, managers, manualworkers, and medical doctors, but the data of all these sub-groups had to be cumulated intoone because of insufficient number of questionnaires per subgroups. As a result, weobtained three basic occupational groups of comparable sample size.

3.2.3.1.1. Onomasiological Types

The data offered in Table 5 indicate that there are differences between the groups ofstudents and teachers on one hand, and the ‘other’ professions, on the other. They mainlyconcern Onomasiological Type 1 where the respective Productivity Rates are 53.86% and57.30%, for the first two informant groups, and much lower in the ‘other’ group (43.25%).This is, naturally, projected onto the situation in Onomasiological Types 2 and 3 where thePRs in the ‘other’ professions are the highest of all. Since the main difference seems to bebetween those who are in education professions and those who are not, the students andeducators have been grouped together in the statistical tests. (OT4 has been left out of thechi-square test because it was zero for all groups.)

The tendency emerging from the data outlined is that while the language speakersbelonging in the education-oriented professions, including education-related major activity(study), tend to form more comprehensive naming units, aimed at maximum explicitnessand accuracy of ‘labeling’ the objects of naming, the speakers belonging in the ‘other’professions prefer morphematically reduced ways of expression (economy of expression)(Type 2), and/or vaguer naming units with broader extension, the meaning of which ismore difficult to predict (Type 3). Thus, in this particular case, we witness a differenttreatment by the representatives of different groups of professions of the above-indicatedconflict between the explicitness of expression and the economy of expression.

26

Education Professions Non-education Professions No. of responses 900 400OT1 497 (55.2%) 173 (43.3%)OT2 193 (21.4%) 106 (26.5%)OT3 193 (21.4%) 110 (27.5%)OT4 0 0 (0%)OT5 17 (1.9%) 11 (2.8%)

Chi-square = 16.089 p = .001 df = 3

Table 5: PR (%) of Onomasiological Types by occupation (native speakers)

3.2.3.1.2. Word-Formation Types

The five most productive WFTs follow the tendencies from the general discussion ofOnomasiological Types. Also here, the PR values of more explicit Word-Formation Typesin the education-oriented groups are higher than those in the non-education group, even ifthe differences are not significant by a chi-square test, and are distributed among theindividual WFTs to give the indicated cumulative effect – the PRs of three-constituentWFTs are generally higher in the education-oriented occupations than in the third group ofinformants. The tendency observed for Onomasiological Types gets the most persuasivesupport from the [Action-Agent] type whose PR in the ‘non-education group’ (25.75%)clearly outscores those for the other two groups (22.61% and 21.35%, respectively).

The largest number of different WFTs has been found in the group of students (16). Theother two groups proposed naming units belonging in 14 different WFTs. This highnumber of different WFTs used and the differences between the individual occupationalgroups provide unequivocal evidence of the validity of the concept of word-formation as‘creativity within productivity constraints’.

Education Non-educationProfessions professions

Object–Action–Agent 283 (34.4%) 130 (32.5%)Action–Agent 199 (24.2%) 103 (25.8%)Theme–Action–Agent 158 (19.2%) 67 (16.8%)Instrument–Action–Agent 97 (11.8%) 38 (9.5%)Location–Action–Agent 86 (10.5%) 32 (8.0%)

chi-square = 2.907 p = 0.5735 df = 4

Table 6: PR (%) of the top five Word-Formation Types by occupations (native speakers)

3.2.3.1.3. Morphological Types

The explicitness-economy conflict and its occupation-based solution at the level ofOnomasiological Types is unambiguously acknowledged at the level of MorphologicalTypes by similar PR differences: while a three-constituent structure [S+S+suffix], in whichall three constituents of the onomasiological structure are morphematically expressed,dominates the education-related groups (47.47% and 50.57%, respectively, versus 41.09%for the ‘other’ group), the highest PR in the ‘other’ professions is achieved by the[S+suffix] structure (42.12% versus 36.59% for students and 33.71% for educators), and

27

the PR of another relatively productive two-constituent structure, [S+S], is also higher inthis group of informants (13.18% versus 11.07% for students and 10.29% for educators).

Education Non-educationprofessions professions

S + S + suffix 438 (51.3%) 164 (41.1%)S + suffix 319 (37.4%) 168 (42.1%)S + S 97 (11.4%) 53 (13.2%)Other 46 (5.1%) 15 (3.6%)

chi-square = 9.252 p = 0.0261 df = 3

Table 7: PR (%) of the most productive Morphological Types by occupations (native speakers)

3.2.3.1.4. Word-Formation Rules

The above-mentioned results are weakly supported by the data of the domain of Word-Formation Rules. In principle, they detail the general results obtained for OnomasiologicalTypes, and therefore the results cannot differ significantly. By implication, the mostproductive WFR for education-related professions in the field examined is (25):

(25) Object–Action–AgentN V -er

i.e. an explicit, three constituent structure both at the onomasiological and theonomatological levels. For non–education professions, this WFR is surpassed by a twoconstituent WFR

(26) Action–AgentV -er

by one response. In the education fields, a slightly higher percentage of the WFRs arebased on Onomasiological Type 1, while the WFRs without morphematic expression of thedetermining constituent of onomasiological structure play a more important role among the‘other’ professions.

Education OtherProfessions professions

Object–Action–Agent (N V –er) 198 (37.5%) 74 (18.5%)Action–Agent (V -er) 157 (29.7%) 75 (18.8%)Theme–Action–Agent (N V –er) 106 (20.1%) 36 (9.0%)Location–Action–Agent (N V-er) 67 (12.7%) 25 (6.25%)

Chi-square = 2.663 p = .4465 df = 3

Table 8 PR (%) of the most productive Word-Formation Rules by occupations (native speakers)

3.2.3.2. Summary

In summarizing the observations based on the experimental data, the following may beconcluded:

28

There is an obvious tendency indicating different strategies in the naming acts in twodifferent groups of language users. While the education-process-related English languageusers incline to those Onomasiological, Word-Formation, and Morphological Types andWord-Formation Rules that are more explicit, thus capturing the objects to be named ina more comprehensive way, the ‘other’ professions prefer brevity of expression, i.e., theyfavour economy of expression, simpler, more general, and therefore, less definite namingunits. The first tendency is interpretation-friendly, because the meaning of a more explicitnaming structure is more easily interpretable and predictable. The latter tendency favoursthe opposite universal feature of language, i.e., the effort for the maximum possibleeconomy of speech to the detriment of clarity of expression.

3.2.3.3. Perception of ‘Unproductivity’

The data indicate that the perception of ‘unproductivity’ among native speakers in generalis fairly strong. While almost all informants gave relevant examples of use of unproductivecoinages in sentences they prevailingly reject these words as extremely unlikely. Inparticular, out of 708 responses, 397 (56.1%) fall within the ‘extremely unlikely’ class ofanswers, and 176 responses (24.9%) in the class of ‘somewhat unlikely’. Thus, the samplewords are considered to be unlikely to over 80%. Yet, there are some differences amongthe individual naming units, with the greatest number of ‘likely-oriented’ responses beingfor sleepable – the only naming unit in this sample, for which there is more ‘somewhatunlikely’ votes than ‘extremely unlikely’ ones. In addition, the number of ‘likelys’ is fairlyhigh. The great majority of the ‘likelys’ are connected with the meaning of ‘apt forsleeping’, mostly with the Location argument, in some cases also with the Temporalargument, for instance, ‘That bed looks very sleepable’, ‘This noise maked the room farfrom sleepable’, ‘The bears about to go to hibernation could be considered in a sleepablestate’, etc.

The data indicate that the constraint, in particular, the subcategorization restrictionpermitting the suffix -able to combine with transitive verbs only, does not seem to be sostrongly anchored in the minds of language users as the other restrictions covered in ourexperiment.

The differences among the individual occupation groups in terms of their respectiveperception of such naming units are not significant, with the exception of sleepable, inwhich case the ‘extremely unlikely’ votes are distributed with steps by about 10 per cent:18.3% for students, 28.6% for educators, and 39.6% for other professions. With thisnaming unit, the percentage of ‘likely’ responses among the students is extraordinarily high– as much as 28.3%.

In general, the number of ‘very likely’ and especially ‘extremely likely’ responsesapproaches zero in the majority of cases, with the exceptions apparently being related toindividual, idiosyncratic, usually stylistically motivated evaluation of a particular samplenaming unit (as suggested by three of the informants who avoided classifying swimmee andsleepable as ‘extremely unlikely’ but emphasized that they could imagine the use of suchwords in ‘jocular’ context only).

The dominating prevalence of the ‘extremely unlikely’ and ‘somewhat unlikely’ responsessuggests that the informants, irrespective of their occupation, have a strong awareness of‘grammaticality’, hence of the relevant productivity constraints. On the other hand, theexistence of a relatively high number of the ‘likely’ responses acknowledges their feeling

29

for a creative approach to their language. This mainly applies to the group of students whomost readily accept unconventional naming units and break the existing rules. This doesnot seem to be a surprise, and might be accounted for psychologically by the dynamism ofthe young generation compared to the more conservative generation of their parents.

Legend: EU – extremely unlikelySU – somewhat unlikelyL – likelyVL – very likelyEL – extremely likely

Students Educators Other professionsEngroupmentEU 31 (52.5%) 18 (52.9%) 33 (67.3%)SU 19 (32.3%) 13 (38.2%) 11 (22.4%)L 8 (13.6%) 1 (2.9%) 5 (10.2%)VL 1 (1.7%) 1 (2.9%) 0 (0%)EL 0 (0%) 1 (2.9%) 0 (0%)Chi-square = 3.910 p = 0.4183 df = 4 (L, VL, and EL were combined)

ThinnenEU 30 (50.8%) 26 (76.5%) 35 (72.9%)SU 20 (33.9%) 5 (14.7%) 5 (10.4%)L 8 (13.6%) 1 (2.9%) 4 (8.3%)VL 1 (1.7%) 2 (5.9%) 3 (6.3%)EL 0 (0%) 0 (0%) 1 (2.1%)Chi-square = 11.699 p = 0.0197 df = 4 (L, VL, and EL were combined)

SwimmeeEU 32 (53.3%) 24 (70.6%) 35 (74.5%)SU 18 (30.0%) 7 (20.6%) 6 (12.8%)L 9 (15.0%) 2 (5.9%) 4 (8.5%)VL 1 (1.7%) 1 (2.9%) 2 (4.3%)EL 0 (0%) 0 (0%) 0 (0%)Chi-square = 7.192 p = 0.1260 df = 4 (L, VL, and EL were combined)

SleepableEU 11 (18.3%) 10 (28.6%) 19 (39.6%)SU 17 (28.3%) 18 (51.4%) 13 (27.1%)L 17 (28.3%) 5 (14.3%) 14 (29.2%)VL 12 (20.0%) 2 (5.7%) 2 (4.2%)EL 3 (5.0%) 0 (0%) 0 (0%)Chi-square = 15.608 p = 0.0035 df = 4 (L, VL, and EL were combined)Chi-square = 21.629 p = 0.0014 df = 6 (VL and EL were combined)

30

SatisfactorityEU 35 (58.3%) 26 (78.8%) 32 (66.7%)SU 11 (18.3%) 4 (12.1%) 9 (18.8%)L 8 (13.3%) 2 (6.1%) 3 (6.3%)VL 4 (6.7%) 1 (3.0%) 3 (6.3%)EL 2 (3.3%) 0 (0%) 1 (2.1%)Chi-square = 4.758 p = 0.3130 df = 4 (L, VL, and EL were combined)

Table 9: Perception of ‘unproductivity’ by native speakers19

3.2.4. Influence of Education

In reference to various incorrect interpretations of compounds, such as house-bird glass,Gleitman/Gleitman (1970) relate their misinterpretation to the educational level oflanguage users. Their informants fell within three different educational groups: (a) graduatestudents and PhD’s in various fields; (b) undergraduates and college graduates; and (c)secretaries with high school degrees. In many cases, their informants from the group ofsecretaries proposed various ‘unacceptable’ readings which corresponded to the compoundglass house-bird, glass bird-house, or a paraphrase like a house-bird made of glass (incontrast to PhD informants who avoided such errors). The analysis of their research resultsmade Gleitman & Gleitman conclude that there were “very large and consistent differencesamong these subjects of differing educational background” (1970: 117) and that “[t]he lesseducated groups make more errors, and to a significant extent make different errors thanthe most-educated group” (ibid. 128). While the research of the Gleitmans concerns thepredictability of meaning, i.e., the interpreter’s pole rather than the coiner’s pole, itindicates that the level of education may play a role in word-formation, in general, and inthe productivity of word-formation, in particular. No wonder, productivity of Word-Formation Rules appears to be one of the factors influencing the predictability of novelcomplex words (cf. Štekauer 2005).

In analyzing the questionnaires, our native speaker informants was divided into threegroups, including those with high school, college, and graduate education. The totalnumber of responses taken into consideration in evaluating the research data was 1,276 forthe category ‘Agents’ nouns.

3.2.4.1. Analysis of the Experimental Data

3.2.4.1.1. Onomasiological Types

The data for the Onomasiological Type 1 show a rising curve in the direction towardshigher education level, though a chi-square test cannot establish significance. TheOnomasiological Type 1 PR of the graduate group is noticeably higher than that of thehigh-school informants. This is compensated for by the higher PRs of the high-school andcollege informants for the other three Onomasiological Types. The highest PR in theOnomasiological Type 3 (one without the morphematic expression of the Actionalconstituent) is attributable to the informants with the lowest education-level, with the PRcurve falling down towards the higher-educated speakers. The data indicate the preferenceof higher educated people for explicit way of expression, and the preference for moregeneral way of expression in the lower educated language users.

19 Note: Not all informants provided answers to all individual tasks. Hence the numbers may differ.

31

High school College GraduateNo. of responses 245 715 316OT1 119 (48.6%) 358 (50.1) 176 (55.7%)OT2 57 (23.3%) 171 (23.9) 69 (21.8%)OT3 63 (25.7%) 169 (23.6) 68 (21.5%)OT4 0 (0%) 0 (0 %) 0 (0%)OT5 6 (2.5%) 17 (2.37% 3 (1.0%)

Chi-square = 5.655 p = 0.4629 df = 6 (OT4 was excluded from calculations)

Table 10: PR (%) of Onomasiological Types by education (native speakers)

3.2.4.1.2. Word-Formation Types

Given the previous data, the downward-leading PR curve in the direction towards highereducation for the [Action-Agent] WFT was expected. As with the majority of other tables,the ranking of the individual WFTs is the same for all three groups of informants.

High school College GraduateNo. of responses 245 715 316Object–Action–Agent 77 (31.5%) 212 (29.7%) 102 (32.3%)Action–Agent 70 (28.5%) 188 (26.3%) 71 (22.4%) Theme–Action–Agent 47 (19.2%) 125 (17.6%) 49 (15.4%)Instrument–Action–Agent 24 (9.6%) 75 (10.5%) 33 (10.6%) Location–Action–Agent 16 (6.5%) 44 (8.1%) 33 (10.6%)Other 11 (4.6%) 57 (7.9%) 28 (8.8%)

Chi-square = 13.378 p = 0.2032 df =10

Table 11: PR (%) of top 5 Word-Formation Types by education (native speakers)

3.2.4.1.3. Morphological Types

The data for the [S + S + Suffix] structure in Table 12 acknowledge the growingimportance of a more complex morphematic representation of complex onomasiologicalstructure, i.e., more precise expression with the growing education of language users (eventhough the differences in PR between the high-school and college informants are minimal).

High school College GraduateNo. of responses 245 715 316S + S + suffix 107 (43.7%) 317 (44.4%) 151 (47.9%)S + suffix 105 (42.9%) 290 (40.5%) 119 (37.8%)S + S 24 (9.8%) 81 (11.2%) 25 (7.9%)

Chi-square = 8.690 p = 0.1917 df = 6

Table 12: PR (%) of the most productive Morphological Types by education (native speakers)

3.2.4.1.4. Word-Formation Rules

The level of Word-Formation Rules used for the coining of Agent names seemingly doesnot bear out the different naming strategies of the speakers of different education levels;

32

this bears on the data obtained for the top PR WFRs. It goes without saying that thedifferent strategies revealed at the levels of Word-Formation Types and MorphologicalTypes must find their mapping also in the domain of WFRs. The differences, however, arenot so conspicuous, because they are scattered among the numerous low PR WFRs. Anindicator of such low PR range differences is the last WFT given in Table 13, showing aPR gap of about 3% between the graduate speakers, on one hand, and the other two groups,on the other. The PR gap of 3% in the low predictability level range is striking.

High school College GraduateNumber of responses 245 715 316Object–Action–Agent N V -er 56 (22.7%) 145 (20.3%) 71 (22.4%)Action–Agent V -er 50 (20.4%) 146 (20.4%) 63 (19.9%)Theme–Action–Agent N V -er 29 (11.9%) 78 (10.8%) 27 (8.5%)Location–Action–Agent N V -er 15 (6.2%) 44 (6.2%) 29 (9.1%)

Chi-square = 6.3676 p = 0.6061 df = 8

Table 13: PR (%) of most productive Word-Formation Rules by education

3.2.4.2. Summary

Tables 10 - 13 suggest, albeit weakly, that education seems to exerts influence upon theapproach to word-formation. There is a noticeable inclination of higher educated people tolabel objects of extra-linguistic reality as precisely as possible and, for this purpose, toemploy more extensive naming structures. Lower educated informants demonstrated theirpreference for more ‘economic’ expressions to the detriment of clarity and precision ofnew naming units.

3.2.4.3. Perception of ‘Unproductivity’

High school College GraduateNumber of responses 34 92 34EngroupmentEU 21 (61.8%) 52 (56.5%) 24 (70.6%)SU 11 (32.4%) 28 (30.4%) 8 (23.5%)L 2 (5.9%) 9 (9.8%) 2 (5.9%)VL 0 (0 %) 2 (2.2%) 0 (0%)EL 0 (0%) 1 (1.1%) 0 (0%) Chi-square = 3.358 p = 0.4998 df = 4 (L, VL, and EL were combined)

ThinnenEU 19 (55.9%) 25 (63.0%) 28 (82.4%)SU 10 (29.4%) 22 (23.9%) 3 (8.8%)L 2 (5.9%) 12 (13.0%) 1 (2.9%)VL 3 (8.8%) 0 (0%) 2 (5.9%)EL 0 (0%) 0 (0%) 0 (0%)Chi-square = 5.7480 p = 0.2187 df = 4 (L, VL, and EL were combined)

33

SwimmeeEU 21 (61.8%) 57 (62.0%) 26 (76.5%)SU 6 (17.7%) 23 (25.0%) 5 (14.7%)L 6 (17.7%) 9 (9.8%) 1 (2.9%)VL 1 (2.9%) 2 (2.2%) 2 (5.9%)EL 0 (0%) 1 (1.1%) 0 (0%)Chi-square = 4.1656 p = 0.3840 df = 4 (L, VL, and EL were combined)

SleepableEU 9 (26.5%) 25 (27.2%) 15 (44.1%)SU 10 (29.4%) 27 (29.4%) 12 (35.3%)L 9 (26.5%) 27 (29.4%) 5 (14.7%)VL 5 (14.7%) 11 (12.0%) 1 (3.0%)EL 1 (3.0%) 2 (2.2%) 1 (3.0%)Chi-square = 6.556 p = 0.1613 df = 4 (L, VL, and EL were combined)

SatisfactorityEU 20 (58.8%) 59 (64.1%) 30 (88.2%)SU 6 (17.7%) 14 (15.2%) 3 (8.8%)L 5 (14.7%) 9 (9.8%) 0 (0%)VL 3 (8.8%) 6 (6.5%) 1 (2.9%)EL 0 (0.0%) 4 (4.4%) 0 (0%)Chi-square = 9.040 p = 0.0601 df = 4 (L, VL, and EL were combined)

Table 14: Perception of ‘unproductivity’ by native speakers

While none of these distributions can be shown to be significant by a chi-square test(though satisfactority comes close), the data still show some suggestive trends.Consistently, the ‘extremely unlikely’ assessment is higher for the ‘graduate’ group. Thisholds true of all five ‘unproductively’ formed sample naming units. These results suggestthat people with more education make stronger judgments of grammaticality. People withmore education could very likely be more committed to notions of correctness, includingfor Word-Formation Rules, and thus are more reluctant to accept words that appear‘ungrammatical.’ The differences between college and graduate informants follow thesame trend, though the differences are smaller. In general, the negative attitude to theungrammaticality of coinages grows with the education of native speakers, with the majorleap in this attitude characterizes the graduate group of the informants.

To conclude, the unproductivity experiment data provide us with another piece of evidenceof education-related differences in the naming strategies.

3.2.5. Influence of Other Languages

3.2.5.1. General

The sample of informants encompasses 109 speakers20 presently living in an Englishspeaking country, but born to parents coming from non-English speaking countries. Thedata acquired from questionnaires indicate that while their parents are fluent in their

20 The actual number of informants was 112. However, the group of polysynthetic language speakers wastoo small (3 informants). By implication, these informants were not included in our analysis.

34

mother tongue none of them can speak English with proficiency corresponding to a nativespeaker.

The expected total number of questionnaire responses (4 tasks with 5 sub-tasks eachaccounts for 20 responses per informant) provided by 109 informants is 2180. In fact, theyprovided 1012 relevant responses for the category of Agent nouns. The reasons why someof the responses have had to be excluded from the analysis are analogical to those in thenative-speaker group of informants.

The sample includes some sociolinguistically complicated cases. Thus, for example,although an informant was born in Holland, his/her native language is Vietnamese andhis/her parents were born in China and Vietnam. Another case is an informant born inMoldova with Romanian as a native language. One of his/her parents was born in Germanywith Russian as a native language, and the other parent was born in Moldova withRomanian as a native language. Since the language most frequently spoken at home is alsoRomanian he is analyzed in the group of analytic/isolating languages. The same criterion isapplied to an informant born in Switzerland one of whose parents was born in France withFrench as a native language.

The informants were divided into four groups based on the morphological typology oflanguages. It is generally known that there are hardly any morphologically pure languages.Given the focus of our experimental research on Agentive nouns, in classifying thelanguages the most important criterion was the prevailing morphological features of nouns.As a result we obtained the following groups

SYNTHETIC/AGLUTTINATING (19 informants) – Korean, Japanese, Finnish,Hungarian, Estonian, Armenian, Swedish, Norwegian, Tagalog, TongaSYNTHETIC/FUSIONAL (17 informants) – German, Slovak, Russian, Polish, Croatian,Czech, Ukrainian, Arabic, Urdu ANALYTIC/ISOLATING (73 informants) –French, Portuguese, Romanian, Italian,Spanish, Dutch, Chinese, Bulgarian, Bangla, Samoan, Creol, Afrikaans, Mandarin,Chinese, Vietnamese, CantonesePOLYSYNTHETIC (3 informants) – Indonesian, Laotian, Hmong

3.2.5.2. Analysis of the Experimental Data

3.2.5.2.1. Productivity of Onomasiological Types

The most noticeable difference between native speakers and non-native speakers asdemonstrated in Table 15 is the respective roles played by Onomasiological Types 1 and 3in these two groups of informants. With the other three Onomasiological Types featuringalmost identical productivity, the PR for the Onomasiological Type 1 is higher by about 4% in the non-native group, and the PR of the Onomasiological Type 3 is lower by the samevalue in the same group of informants.

We may surmise that one of the reasons for this difference is as follows: since theinformants, falling within the non-native group, do not master English as fluently as nativespeakers (levels B and lower in the questionnaire) their linguistic uncertainty makes themtry very hard in the naming act to make their ‘products’ as comprehensible as possible, andtherefore, most explicit. For this reason, they prefer Onomasiological Type 1. Obviously,this is a possible psychological motivation behind this preference. The second reason may

35

be connected with the structural characteristics of the non-English languages that wereshaping the linguistic behaviour of the informants in the past.

Native speakers Non-native speakers No. of responses 1300 1012OT1 670 (51.5%) 561 (55.4%)OT2 299 (23.0%) 236 (23.3.1%)OT3 303 (23.3%) 193 (19.1%)OT4 0 (0%) 0 (0%)OT5 28 (2.2%) 22 (2.2%)Chi-square 6.4094 p = 0.0933 df = 3 (OT4 was excluded from calculations)

Table 15: PR (%) of Onomasiological Types (non-native speakers)

3.2.5.2.2. Productivity of Word-Formation Types

The differences discussed in the previous section cannot be, for obvious reasons,manifested at the level of WFTs. In spite of this fact, the agreement of the respective PRs issurprisingly high. While the top WFT ranking agreement was expected the PR differencesare extraordinarily small (for the top five WFTs in succession: 0.19; 0.26; 0.61; 1.37; and1.77 %, respectively).

Native speakers Non-native speakersNo. of responses 1300 1012Object–Action–Agent 416 (32.0%) 322 (31.8%)Action–Agent 304 (23.4%) 234 (23.1%)Theme–Action–Agent 225 (17.3%) 169 (16.7%)Instrument–Action–Agent 135 (10.4%) 119 (11.8%)Location–Action–Agent 118 (9.1%) 74 (7.3%)

Chi-square = 4.6546 p = 0.4594 df = 5

Table 16 PR (%) of the top five Word-Formation Types (non-native speakers)

3.2.5.2.3. Productivity of Morphological Types

The suffix-based Morphological Types of [S+S+suffix] and [S + Suffix] for Agent namesis understandable with respect to the large number of Agentive suffixes in English. Whenthe central suffix-based Agent types are added up, they represent 84.08% for the nativegroup and 77.67% for the non-native group. What makes the two groups of informantsdifferent is the much stronger role of the [S + suffix] MT in the native group of speakers(37.41% compared to 28.66% in the non-native group), and, on the other hand, a slightlyhigher PR of the [S+S+suffix] MT in the non-native group. These data correspond with theobservations concerning the productivity of Onomasiological Types.

36

Native speakers Non-native speakersNo. of Responses 1300 1012S + S + suffix 607 (46.7%) 496 (49.01%)S + suffix 486 (37.4%) 290 (28.66%)S + S 147 (11.3%) 163 (16.11%)

Chi-square = 26.1042 p < .0001 df = 3

Table 17: PR (%) of Morphological Types (non-native speakers)

3.2.5.2.4. Productivity of Word-Formation Rules

The results for the Onomasiological Types are also mapped onto the level of Word-Formation Rules. First, while in the group of native speakers there is one WFR in the ‘top5 chart’ in which the determined constituent of the onomasiological mark is not expressed,there is no such WFR among the top five in the non-native group. Second, the strongposition of OT2 is supported by two WFRs among the top 5, in which the determiningconstituent is not expressed (ranks 2 and 5) as opposed to only one such WFR in the nativespeaker group.

Native speakers Non-native speakers

No. of Responses 1300 1012Object–Action–AgentN V -er 272 (20.9%) 222 (21.9%)Action–Agent V -er 232 (17.9%) 159 (15.7%)Theme–Action–Agent N V -er 138 (10.6%) 121 (11.96%)Location–Action–Agent N V -er 92 (7.1%) 65 (6.4%)Action–Agent V N 35 (2.7%) 57 (5.6%)

Chi-square = 16.3394 p = 0.0059 df = 5

Table 18: PR (%) of Word-Formation Rules (non-native speakers)

3.2.5.2.5. Summary

The comparison of the naming behaviour of the native and the non-native informants hasshown considerable and significant differences. Their naming strategies appear to differprimarily in the non-native group of speakers laying much greater emphasis on the‘accuracy of naming’, which implies explicitness especially in relation to the determinedconstituent of onomasiological mark.

3.2.5.3. Comparison of Three Cohorts of Influencing Languages

Given the non-existence of a word-formation typology of languages, the ‘background’languages were divided into three groups, based on the morphological typology of noun,that is, the synthetic/agglutinating, synthetic/fusional, and analytic/isolating types. The

37

fourth group, the polysynthetic type, was represented by only three informants, representingthree languages (Indonesian, Laotian and Hmong), and therefore, it was not included in theanalysis.21

3.2.5.3.1. Onomasiological Types

An overview of the results for Onomasiological Types is given in Table 19.

Native speakers Non-native linguistic backgroundAgglutinative Fusional Analytic

No. of responses 1300 180 161 671

OT1 670 (51.5%) 109 (60.56%) 94 (58.39%) 358 (53.35%)OT2 299 (23.0%) 38 (21.10%) 38 (23.60%) 160 (23.85%)OT3 303 (23.3%) 28 (15.56%) 28 (17.40%) 137 (20.42%)OT4 0 (0%) 0 (0%) 0 (0%) 0 (0%)OT5 28 (2.2%) 5 (2.78%) 1 (0.62%) 16 (2.38%)

Chi-square = 12.6218 p = 0.1804 df = 9 (native v. all)Chi-square = 7.21461 p = 0.0653 df = 3 (native v. agglutinative)Chi-square = 5.21562 p = 0.1566 df = 3 (native v. fusional)Chi-square = 2.17689 p = 0.5365 df = 3 (native v. analytic)

Table 19: PR (%) of Onomasiological Types by language background

A crucial observation following from Table 19 is that the hypothesis of the influence of theinfluence of language background seems to have been confirmed. English is predominantlyan analytic language and therefore the results obtained from native speakers should beclosest to those obtained from the isolating/analytic group of background languages. Thedata seem to suggest this, though the differences don’t rise to statistical significance. Thereis a striking similarity between the naming tendencies in these two groups of informants,while the agglutinative and the fusional background languages deviate from the ‘native’data in a noticeable way, as reflected in their lower p-values in the chi-square test. Thisprimarily concerns the role played by Onomasiological Types 1 and 3. Furthermore, it is nosurprise that the agglutinative group’s PR for OT1 is the highest of all. This may beexplained – in addition to the psychological reasons relevant to all three groups of non-native language background – by the morphological characteristics of agglutinativelanguages, aiming at expressing complex morphological meanings within one word.

3.2.5.3.2. Word-Formation Types

An overview of the results for Word-Formation Types is given in Table 20.

21 The classification of the languages in terms of the morphology of noun was based onKrupa/Genzor/Drozdík (1983), Comrie (1981), and the Internet sources, http://www.paul-raedle.de/vtrain/db-xx-info.htm, http://www.geocities.com/indoeurop/tree/balk/armenian.html, http://www.cambridge.org/uk/catalogue/catalogue.asp?isbn=0521573785&ss=exc, http://www.linguistics.emory.edu/POLYGLOT/morphology.html, and on personal communication. Therefore, we wish to express ourgratitude for help to Jan Don, A. Olofsson, S. Valera, and M. Volpe.

38

Native speakers Non-native linguistic background

Agglutinative Fusional AnalyticNo. of responses 1300 180 161 671

Object-Action-Agent 416 (32.0%) 62 (34.4%) 50 (31.1%) 210 (31.3%)Action-Agent 304 (23.4%) 38 (21.1%) 38 (23.6%) 158 (23.6%)Theme-Action-Agent 225 (17.3%) 26 (14.4%) 27 (16.8%) 116 (17.3%)Instrument–Action–Agent 135 (10.4%) 22 (12.22) 20 (12.42) 77 (11.48)Location-Action-Agent 118 (9.1%) 13 (7.22) 11 (6.83) 50 (7.46)

Chi-square = 11.201 p = 0.7382 df = 15 (native v. all)Chi-square = 2.6524 p = 0.7533 df = 5 (native v. agglutinative)Chi-square = 3.5797 p = 0.6113 df = 5 (native v. fusional)Chi-square = 6.8404 p = 0.2327 df = 5 (native v. analytic)

Table 20: PR (%) of Word-Formation Types by language background

Table 20 gives support to the observations given in 3.2.5.3.1. In each of the top five Word-Formation Types the PRs of native speakers and the ‘analytic’ language background groupof informants are closer to each other than the results obtained from the other two groups,even if the differences between the PRs are small in general. Nonetheless, the ‘native-analytic’ comparison features extraordinarily small differences: 0.7; 0.17; 0.03; 1.10; and1.62 respectively, for the first five WF Types.

3.2.5.3.3. Morphological Types

An overview of the results for Morphological Types is given in Table 21.

Native speakers Non-native linguistic backgroundNative Agglutinative Fusional Analytic

No. of Responses 1300 180 161 671

S+S+suffix 607 (46.7%) 99 (55.0%) 85 (52.8%) 312 (46.5%)S+suffix 486 (37.4%) 47 (26.1%) 41 (25.5%) 202 (30.1%)S+S 147 (11.3%) 22 (12.2%) 29 (18.0%) 112 (16.7%)Other 60 (4.6%) 4 (2.2%) 5 (3.1%) 25 (3.7%)

Chi-square = 32.5464 p = 0.0001 df = 9 (agglutinative v. all)Chi-square = 10.107 p = 0.0176 df = 3 (agglutinative v. native)Chi-square = 2.1152 p = 0.5488 df = 3 (agglutinative v. fusional)Chi-square = 5.6579 p = 0.1294 df = 3 (agglutinative v. analytic)

Table 21: PR (%) of Morphological Types by language background Given the prevailing word-formation tendencies in the languages under evaluation, onemight, in general, expect major differences in Morphological Types and Word-FormationRules. This follows from the purely formal nature of the traditional classification of word-formation processes. Thus, we might expect that the share of the suffix-based types andrules in agglutinative languages will be higher than that in the native group of speakers and

39

in the other two groups of background language. These expectations have been confirmedto a considerable degree, especially with regard to the [S+S+suffix] Morhpological Typewhere the dominance of the agglutinative background is dominant, especially with regardto the native speaker and the isolating background groups of informants. The onlyunexpected outcome is an even lower PR of the [S+S] MT in the native speaker group thanthe PR of the same type in the agglutinative group. A remarkable parallel between theProductivity Rates of the suffixed MTs in the native and the ‘isolating background’ groupscan also be traced here, with the exception of the [S+S] type.

3.2.5.3.4. Word-Formation Rules

An overview of the results for Word-Formation Rules is given in Table 22.

Native speakers Non-native linguistic backgroundAgglutinative Fusional Analytic

No. of Responses 1300 180 161 671

Object–Action–Agent N V -er 272 (20.9%) 49 (27.2%) 36 (22.4%) 137 (20.4%)Action–Agent V -er 232 (17.9%) 26 (14.4%) 24 (14.9%) 109 (16.3%)Theme–Action–Agent N V -er 138 (10.6%) 19 (10.6%) 22 (13.7%) 80 (11.9%)Location–Action–Agent N V -er 92 (7.1%) 12 (6.7%) 11 (6.8%) 42 (6.3%)Action–Agent V N 35 (2.7%) 8 (4.4%) 12 (7.4%) 37 (5.5%)

Chi-square = 23.5232 p = 0.0736 df = 15 (native v. all)Chi-square = 6.3357 p = 0.2749 df = 5 (native v. agglutinative)Chi-square = 13.4806 p = 0.0192 df = 5 (native v. fusional)Chi-square = 11.6355 p = 0.0401 df = 5 (native vs. analytic)

Table 22: PR (%) of Word-Formation Rules by language background

Table 22 also demonstrates a coincidence between the native speakers and those with theanalytic language background. The PRs of the most productive WF Rule in the two groupsof informants are almost identical, significantly differing from the agglutinatingbackground PR, and the same situation may be observed for the second most productiveWF Rule. For other WFRs the differences between the individual groups of informants areminimal.

3.2.5.4. Comparison of the ‘Unproductivity’ Results

3.2.5.4.1. Native vs. Non-native Speakers (as a Whole)

If we concluded in Sections 3.2.3.3 that the perception of unproductivity in native speakersis very strong Table 23 shows us that a similar statement is applicable to the non-nativeinformants. In spite of this general conclusion, there are certain differences between thetwo groups of speakers of English. While the share of the ‘extremely unlikely’ responses in

40

the native group is 56.1%, in the group of non-native speakers it is less (50.0%) whichindicates that the pressure of productive WF rules is perceived by native speakers a littlestronger. This tendency has been borne out in three of the five ‘unproductively’ coinednaming units. Two gaps are significant: almost 16% for satisfactority and over 12% forengroupment. This difference is mostly compensated at the next lower level, the level of‘somewhat unlikely’ answers. In one case (thinnen) we might speak of a draw because thepercentages were almost identical (64.54% vs. 64.76%). The non-native speakersmanifested about 3% higher distaste for sleepable.

The non-native speakers feature higher percentages at the medium assessment level, i.e., atthe level of the ‘likely’ answers. The biggest assessment gap at this level is observed forsatisfactority (10%); in two other cases, swimmee and sleepable, the gap is about 5% infavour of the non-native speakers.

The results are not very conclusive in one or the other direction. What may be assumedbased on them is that native speakers are slightly more ‘aware’ of the productive WFprocesses. On the other hand, the differences are not significant. In both groups ofrespondents, we can observe certain will to creative ‘experimentation’ which depends onthe nature of the constraint violated. Table 23 gives a comparison of the two groups ofinformants for the individual ‘non-words’.

Native speakers Non-native speakersNumber of responses 142 109

EngroupmentEU 82 57.8%) 51 (46.8%)SU 43 (30.3%) 41 (37.6%)L 14 (9.9%) 9 (8.3%)VL 2 (1.4%) 6 (5.5%)EL 0 (0.0%) 2 (1.8%)Chi-square = 3.2609 p = 0.20 df = 2

ThinnenEU 92 (64.5%) 72 (66.1%)SU 30 (21.3%) 26 (23.9%)L 13 (9.2%) 9 (8.3%)VL 6 (4.3%) 1 (0.9%)EL 1 (0.7%) 1 (0.9%)Chi-square = 1.0166 p = 0.60 df = 2

SwimmeeEU 91 (64.50%) 59 (54.1%)SU 31 (22.00%) 32 (29.4%)L 15 (10.60%) 14 (12.8%)VL 4 (2.80%) 4 (3.7%)EL 0 (0%) 0 (0.0%)Chi-square = 2.8198 p = 0.24 df = 2

41

SleepableEU 40 (28.00%) 36 (33.0%)SU 48 (33.60%) 35 (32.1%)L 36 (25.20%) 30 (27.5%)VL 16 (11.20%) 7 (6.4%)EL 3 (2.10%) 1 (0.9%)Chi-square = 0.7811 p = 0.68 df = 2

SatisfactorityEU 93 (66.00%) 58 (53.2%)SU 24 (17.00%) 23 (21.1%)L 13 (9.20%) 17 (15.6%)VL 8 (5.70%) 7 (6.4%)EL 3 (2.10%) 4 (3.7%)Chi-square = 4.4179 p = 0.11 df = 2

Table 23: Perception of ‘unproductivity’: Native vs. non-native informants

3.2.5.4.2. Non-Native Speakers (Individual Types)

As for the internal structure of the non-native informants, a clearly highest resistance tounproductively coined naming units is exercised by those with a fusional languagebackground, much higher than the other two groups of informants. With the exception ofsatisfactority, the differences between the individual groups of informants are very high.For example, thinnen, the difference between agglutinative and the fusional groups is over40%.

We have no explanation for these results. By all accounts, however, the acceptability/non-acceptability of a naming unit coined by violating a restriction on productivity is notinfluenced by the type of a background language.

Native Non-nativeAgglutinative Fusional Analytic

Number of responses 142 (19) (17) (73)

EngroupmentEU 82 (57.8%) 9 (47.4%) 10 (58.8%) 32 (43.8%)SU 43 (30.3%) 7 (36.8%) 5 (29.4%) 29 (39.7%)L 14 (9.9%) 2 (10.5%) 1 (5.9%) 6 (8.2%)VL 2 (1.4%) 1 (5.3%) 0 (0%) 5 (6.9%)EL 0 (0%) 0 (0%) 1 (5.9%) 1 (1.4%)Chi-square = 4.5140 p = .61 df = 6 (native vs. all types; L, VL, and EL combined)Chi-square = 0.8366 p = .66 df = 2 (native vs. agglutinative; ; L, VL, and EL combined)Chi-square = 0.0094 p = .99 df = 2 (native vs. fusional; ; L, VL, and EL combined)Chi-square = 4.0221 p = .13 df = 2 (native vs. analytic; ; L, VL, and EL combined)

42

ThinnenEU 92 (64.5%) 9 (47.4%) 15 (88.2%) 48 (65.8%)SU 30 (21.3%) 7 (36.8%) 1 (5.9%) 18 (24.7%)L 13 (9.2%) 3 (15.8%) 1 (5.9%) 5 (6.9%)VL 6 (4.3%) 0 (0%) 0 ( 0%) 1 (1.4%)EL 1 (0.7%) 0 (0%) 0 (0%) 1 (1.4%)Chi-square = 7.9900 p = .24 df = 6 (native vs. all types; L, VL, and EL combined)Chi-square = 2.6456 p = .27 df = 2 (native vs. agglutinative; ; L, VL, and EL combined)Chi-square = 3.8233 p = .15 df = 2 (native vs. fusional; ; L, VL, and EL combined)Chi-square = 1.0520 p = .59 df = 2 (native vs. analytic; ; L, VL, and EL combined)

SwimmeeEU 91 (64.5%) 11 (57.9%) 14 (82.4%) 34 (46.6%)SU 31 (22.0%) 3 (15.8%) 3 (17.7%) 26 (35.6%)L 15 (10.6%) 5 (26.3%) 0 (0%) 9 (12.3%)VL 4 (2.8%) 0 (0%) 0 (0%) 4 (5.5%)EL 0 (0%) 0 (0%) 0 (0%) 0 (0%)Chi-square = 13.7123 p = .03 df = 6 (native vs. all types; L, VL, and EL combined)Chi-square = 7.12163 p = .03 df = 2 (native vs. agglutinative; ; L, VL, and EL combined)Chi-square = 21.0597 p < .0001 df = 2 (native vs. fusional; ; L, VL, and EL combined)Chi-square = 11.4897 p = .0031 df = 2 (native vs. analytic; ; L, VL, and EL combined)

SleepableEU 40 (28.0%) 7 (36.8%) 7 (41.2%) 22 (30.1%)SU 48 (33.6%) 5 (26.3%) 7 (41.2%) 23 (31.5%)L 36 (25.2%) 6 (31.6%) 3 (17.7%) 21 (28.8%)VL 16 (11.2%) 0 (0%) 0 (0%) 7 (9.6%)EL 3 (2.1%) 1 (5.3%) 0 (0%) 0 (0%)Chi-square = 3.7022 p = .7169 df = 6 (native vs. all types; L, VL, and EL combined)Chi-square = 0.7358 p = .6921 df = 2 (native vs. agglutinative; ; L, VL, and ELcombined)Chi-square = 2.9737 p = .2260 df = 2 (native vs. fusional; ; L, VL, and EL combined)Chi-square = 0.1414 p = .9317 df = 2 (native vs. analytic; ; L, VL, and EL combined)

SatisfactorityEU 93 (66.0%) 10 (52.6%) 9 (52.9%) 39 (53.4%)SU 24 (17.0%) 5 (26.3%) 5 (29.4%) 13 (17.8%)L 13 ( 9.2%) 4 (21.1%) 2 (11.8%) 11 (15.1%)VL 8 (5.7%) 0 (0%) 0 (0%) 7 (9.6%)EL 3 (2.1%) 0 (0%) 1 (5.9%) 3 (4.1%)Chi-square = 6.7976 p = .3399 df = 6 (native vs. all types; L, VL, and EL combined)Chi-square = 1.4154 p = .4927 df = 2 (native vs. agglutinative; ; L, VL, and ELcombined)Chi-square = 1.8914 p = .3884 df = 2 (native vs. fusional; ; L, VL, and EL combined)Chi-square = 4.3977 p = .1109 df = 2 (native vs. analytic; ; L, VL, and EL combined)

Table 24: Perception of ‘unproductivity’ (%): Native vs. Non-native informants

43

3.2.6. Additional Remarks

3.2.6.1. Double Formal Indication of Agent – Redundancy in Word-Formation

Strangely, the questions of word-formation redundancy have been paid little attention inthe relevant literature.22 The redundancy phenomenon in word-formation is closely relatedto one of the central points of discussion of our research, in particular, the conflict betweenthe explicitness of expression and the economy of expression.

Double indication of a single conceptual category runs counter to the very notion of theeconomy of expression, and counter to a linguistic notion that there should be one to onecorrespondence of conceptual and formal categories in word-formation. Certainly, the stateof isomorphy is an ideal one in morphology in general. In practice, there are a number ofcases where a single conceptual category is represented in a language by a number ofallomorphs/synonymous morphemes (Agent nouns themselves are a case in point). What israre in the English language, however, is the doubling of the same formal means within onenaming unit, which introduces redundancy. This is captured by Lieber’s RedundancyRestriction (2004: 161):

(27) The Redundancy PrincipleAffixed do not add semantic content that is already available within a baseword (simplex or derived).

Therefore, it may be surprising to find relatively numerous cases of this sort in ourresearch, including butter-inner, hanger-onner, butter-innist; weberer, shoe-tier-upper,grass-cutter-upper, on the one hand, and bird-fisherman, shoe-tierman, hangerman, on theother. The former type, characteristic of nonstandard and informal language and casual,perhaps jocular, speech , may be accounted for by language users feeling uncomfortableputting Agentive suffix on the particle of phrasal verbs. Yet, at the same time, theyrecognize that agentive suffixes go at the end, so they end up putting a suffix on theparticle as well as the verb. With reference to the phrasal verb butt in, the Americanmember of the evaluative team finds butter inner more ‘natural’ than butt inner or evenbutter in. This assumption has been experimentally acknowledged: the proportion betweenthe occurrences of butter inner and butter in our research is 17 to 8, with zero occurrenceof butt inner. As for the internal structure of butter inner, students selected it most of alloccupational categories – nine times, which is more than 50%. On the other hand, thesituation with hanger onner is quite opposite. Its three occurrences represent just 50% ofthe occurrences of hanger-on. Butter-innist only occurred once, and its ‘author’ is a femalemanager from Great Britain.

Weberer is a different case because to web is not a phrasal web, and is difficult to explain.We suspect that it is related to double comparatives that show up – somehow speakers donot recognize the first suffix and end up putting another suffix on. Let us recall the fairlyrecent movie called Dumb and Dumberer which is a sequel to Dumb and Dumber. Thisnaming unit only occurred once, and was proposed by a female informant in the category of‘Management’. The informant lives in Great Britain.23

22 Important exceptions to this rule are Plag (1999) and Lieber (2004).23 All in all, the above-given cases are not mere experimental oddities, which is borne out by the existence of

established words. Lieber (2004: 164) refers to OED citing a number of similar examples, such ascheckerist, consumerist, collegianer, musicianer, etc. In addition, the Agent-related redundancy is not theonly type of redundancy in English. A much more frequent type are the -ic-al adjectives. For further

44

3.2.6.2. Suffix -sky (-insky)

This suffix occurred in the naming unit buttinsky, apparently attached to the verbal basebutt-in in the meaning ‘someone who has something to do with V’. It is certainly nota common suffix across varieties of English; rather, it seems to be an Eastern U.S.regionalism. There is definitely something playful or slangish about it. Its origin may besupposed to be in the -sky formative encountered in Polish names. Its connotations seem tobe slightly pejorative, perhaps suggesting someone who is boorish in connection with anitem or an action. It may be more popular among working class, and indeed, in ourexperimental results, three male and one female informants and all self-identified in the‘Manual Work’ category chose this option. These four informants represent a third of allthose identified in the ‘Manual Work’ occupation.

3.2.6.3. Blends

Blends appear to be quite popular with coiners, supporting the economy of expression atthe expense of meaning clarity. The experiment came up with several interesting blends:

Persniskigardener – a blend of gardener and persnickety ‘fastidious, overly attentive todetails, excessively demanding’. Its single occurrence is related to a male manual workerfrom the USA.

Blondoronious – there are two possible interpretations of this naming unit. Either it isa blend combining blond and errorneous, which gives the meaning ‘someone who iserroneous about blonds’. Another possible interpretation is one based on a pretentious (andtherefore playful) suffix.

Blonde-ogynist – a blend of blonde + misogynist, proposed by an American female teacher.This is a fascinating formation, since it suggests that the blonds who are ridiculed arefemale. It also leaves out the part of mysoginist that explicitly marks ‘hatred’ andreinterprets the last part for that.

Laceanomist – a blend of lace + -onomist. –onomist shows up on a number of wordsindicating an expert at a (usually academic or professional) field, such as ‘economist’ or‘agronomist’. This blend appears to be used to attach some prestige, or at least the notionof a profession, to someone who ties shoes.

3.2.6.4. Other Interesting Cases

Car-top boogieborder – this naming unit, proposed by an American male teacher, is usedto denote a person depicted in the picture on car-top, and makes use of figurativeexpression, i.e., someone who rides on the car top as they would a boogie board (a smallsurf board made to be ridden in the prone position, as in the picture).

Anal-lawn maintenance worker; anal grass snipper and anal-retentive – the basic term inthis group, anal-retentive, comes from Freud’s notions of child development. Apparently,in Freud’s thought, the stages of toilet-training can lead some to become too preoccupiedwith structure and order and detail, and this is the general meaning of anal. Thus, for

discussion see Plag (1999) and Lieber (2004).

45

instance, when students think that some English usage rules are too picky, they might saysomething like ‘that rule is just anal’. The first two terms were used by American femaleteachers, the last one by an American natural scientist.

Representor, race representor – the creative aspect is manifested here very clearly; whilethere is a word representative in the meaning ‘a person duly authorized to act or speak foranother or others’, the coiners (an American male young unemployed informant and anAmerican female teacher, respectively) apparently wanted to emphasize the new role ofa person who represents the whole mankind by having recourse to a fully grammaticalcoinage using the suffix -er.

4. Conclusions

1. The research has confirmed the concept of word-formation conceived as creativitywithin productivity constraints. While the effectiveness of ‘productivity constraints’ aremanifested by the types and rules with high Productivity Rates and by the extensivecoincidence of their ranking in the various experimental groups, the word-formation‘creativity’ is borne out by the diversity of the types and rules fulfilling the same functionwithin a particular conceptually defined cluster. The present research gives ample evidencein favour of this approach to productivity, and shows that rather than excluding each other(as traditionally believed) productivity and creativity co-exist.

2. The proposed method of productivity calculation proved to be a feasible tool for anobjective evaluation of the role of the individual types and rules without any unjustifiedpreference for any particular word-formation process (as opposed to the mainstreamaffixation-oriented approaches). This method makes it possible to evaluate the productivityat different levels of generalization, to reflect its different aspects, including the mostgeneral onomasiological level; onomasiological structure (logico-semantic relations);onomatological structure (formal realization of coinages); and the interrelation of theonomasiological and the onomatological structures (established by the Morpheme-to-SemeAssignment principle). Importantly, each of these levels of productivity calculationencompasses any and all of the traditional formally defined word-formation processes.Furthermore, this method makes it possible to avoid the classification problems socharacteristic of the generative approach to word-formation (compounding vs. affixation,bracketing paradoxes) thanks to the fact that all word-formation processes are treated ina consistent onomasiological manner, and therefore, defined on the basis of a single,unifying principle.

As far as the specific targets of our experimental research are concerned the followingconclusions may be drawn:

(a) The conflict between the explicitness of expression and the economy of expression inthe field of Agent names favours the explicitness tendency. Language users tend to makeuse of the types and rules which employ the crucial Actional constituent of theonomasiological, and mainly, onomatological structure. It is for this reason that the mostproductive Onomasiological Type is OT1, the most productive Word-Formation Types are[Object–Action–Agent] and [Action–Agent], the most productive Morphological Typesare [S+S+suffix] and [S+suffix], and the most productive Word-Formation Rule is (28)

(28) Object–Action–Agent N V -er

46

(b) The research has borne out the hypothesis of sociolinguistic conditioning of theindividual acts of word-formation. The analysis of the results by occupation has shownthat there is a tendency indicating different strategies taken by education-related and‘other’ professions in the implementation of naming acts. While the former group havea stronger preference for the explicit types and rules, the latter group more frequentlyfavours the more ‘economic’ solutions. Furthermore, the level of education appears to havesimilar effects: while native speakers with university education prefer more precisenames, lower educated speakers are more frequently driven by the principle ofeconomy of expression.

The influence of language-background seems to be equally important. The preference forthe ‘Action-expressed’ Onomasiological Types among non-native speakers is even muchstronger than with native speakers, especially the role of Onomasiological Type 2 isextremely strong. This is, logically, projected onto the high Productivity Rate of theMorphological Type [S + suffix] in this group of speakers, and the absence among the topfive Word-Formation Rules of a rule in which the determined constituent is not expressed.In general, the naming strategies of the two basic groups of speakers seem to differ becausenon-native speakers seem to lay even greater emphasis on the explicitness ofexpression than native speakers.

The influence of linguistic background plays its role in the naming strategies of non-naming speakers. Although the limited sample of informants with ‘Germanic linguisticbackground’ does not enable us to draw any indisputable conclusions, the agreement of theresults between them and the native English speakers in terms of almost identicalpreference for affixal types is remarkable.

(c) It has been shown that any assessment of the influence of any of the above-discussedsociolinguistic factors must be related to the specific conceptually determined categoryof the cluster (Agent, Patient, Instrument, etc.).

(d) Finally, the research into ‘unproductivity’ has demonstrated that the perception of‘unproductivity’ among both native and non-native speakers in general is strong. On theother hand, the existence of a relatively high number of the ‘likely’ responsesacknowledges their feeling for a creative approach to naming. This mainly applies tothe group of students who most readily accept unconventional naming units and break theexisting rules. This does not seem to be a surprise, and might be accounted forpsychologically by the dynamism of the young generation compared to the moreconservative generation of their parents. Moreover, the share of ‘extremely unlikely’answers in the university-educated informants is much higher than in the lower-educatedgroups, which suggests that the awareness of grammaticality of higher-educatedspeakers is stronger. The tolerance to ‘creativity’ (even the creativity that trespassesgrammaticality) characteristic of the young generation is also typical of speakers withlower education, even if the reasons underlying this fact may partly differ in these twogroups of speakers.

The native–non-native comparison shows that while the both groups demonstrate theawareness of unproductivity, there are some differences between the two groups: thepressure of productive Word-Formation Rules is perceived by native speakers a littlestronger.

47

Pavol Štekauer Don ChapmanSlávka Tomaščíková Dept. of Linguistics and English LanguageŠtefan Franko Brigham Young UniversityDepartment of British and American Provo, UT 84602, USAStudies Faculty of Arts e-mail: don_chapman@byu.eduPrešov University17. novembra 108078 Prešov, Slovakiae-mail: stekpal@unipo.sk

References

Adams, Valerie (2001), Complex Words in English, Harlow: Longman.Allen, Margaret R. (1978), Morphological Investigations, Ph.D. Univ. Connecticut.Anderson, Stephen R. (1982), “Where’s Morphology”, Linguistic Inquiry 13: 571-612.Anderson, Stephen R. (1992), A-Morphous Morphology, Cambridge: Cambridge University Press.Anshen, Frank / Aronoff, Mark (1981), “Morphological Productivity and Phonological Transparency”,

Canadian Journal of Linguistics 1: 63-72.Aronoff, Mark (1976), Word Formation in Generative Grammar, [Linguistic Inquiry Monograph 1],

Cambridge, MA: MIT Press.Aronoff, Mark / Schvaneveldt, Roger W. (1978), “Testing Morphological Productivity”, Annals of the New

York Academy of Sciences 318: 106-114.Aronoff, Mark (1983), “Potential Words, Actual Words, Productivity and Frequency”, in: Preprints of the

Plenary Session Papers of the XIIIth International Congress of Linguists, 163-170, Tokyo.Baayen, Harald (1989), A Corpus-Based Approach to Morphological Productivity: Statistical Analysis and

Psycholinguistic Interpretation, Ph.D. Diss. Free University of Amsterdam.Baayen, Harald (1992), “A Quantitative Approach to Morphological Productivity”, in: Booij, Geert / Marle,

Jaap van (eds.), Yearbook of Morphology 1991, 109-149, Dordrecht: Kluwer.Baayen, Harald (1993), “On Frequency, Transparency, and Productivity”, in: Booij, Geert / Marle, Jaap van

(eds.), Yearbook of Morphology 1992, 181-208, Dordrecht: Kluwer.Baayen, Harald (1994a), “Derivational Productivity and Text Typology”, Journal of Quantitative Linguistics

1: 16-34.Baayen, Harald (1994b), “Productivity in Language Production”, Language and Cognitive Processes 9: 447-

469.Baayen, Harald (1996), “The Effects of Lexical Specialization on the Growth Curve of Vocabulary”,

Computational Linguistics 22: 455-480.Baayen, Harald (1997), “Markedness and Productivity”, in: Dressler, Wolfgang, U. / Prinzhorn, Martin /

Rennison, John R. (eds.), Advances in Morphology, 189-200, Berlin: Mouton de Gruyter.Baayen, Harald / Lieber, Rochelle (1991), “Productivity and English Derivation: A Corpus Based Study”,

Linguistics 29: 801-843.Baayen, Harald / Renouf, Antoinette (1996). “Chronicling The Times: Productive Lexical Innovations in an

English Newspaper”, Language 72: 69-96.Barker, Chris (1998), “Episodic -ee in English: A Thematic Role Constraint on New Word Formation”,

Language 4: 695-727.Bauer, Laurie (1983), English Word-Formation, Cambridge: Cambridge University Press.Bauer, Laurie (2001), Morphological Productivity, [Cambridge Studies in Linguistics 95], Cambridge:

Cambridge University Press.Beard, Robert E. (1995), Lexeme-Morpheme Base Morphology. A General Theory of Inflection and Word

Formation, State University of New York Press.Bolinger, Dwight (1975), Aspects of Language, New York: Harcourt Brace Jovanovich.Botha, Rudolf P. (1984), Morphological Mechanisms: Lexicalist Analysis of Synthetic Compounding,

Oxford: Pergamon Press.Chomsky, Noam (1970), “Remarks on Nominalization”, in: Jacobs, Roderick A. / Rosenbaum, Peter S. (eds.),

Readings in English Transformational Grammar, 184-221, Waltham, MA: Ginn,.Com rie, Bernard (1983), Language Universals and Linguistic Typology, Oxford: Basil Blackwell.Crystal, David (1997), Cambridge Encyclopaedia of the English Language, Cambridge: Cambridge

University Press.

48

Di Sciullo, Anna-Maria / Williams, Edwin (1987), On the Definition of Word, [Linguistic Inquiry Monograph14], Cambridge, MA: MIT Press.

Dokulil, Miloš (1962), Tvoření slov v češtině I. Teorie odvozování slov, Praha: NakladatelstvíČeskoslovenské akademie věd.

Dressler, Wolfgang U. / Ladányi, Mária (2000), “Productivity in Word Formation (WF): A MorphologicalApproach”, Acta Linguistica Hungarica 47: 103-144.

Fabb, Nigel (1988), “English Suffixation Is Constrained Only by Selectional Restrictions”, Natural Languageand Linguistic Theory 6: 527-539.

Geeraerts, Dirk (1983), “Reclassifying semantic change”. Quaderni di semantica 4: 217-240.Gleitman, Lila R. / Gleitman, Henry (1970), Phrase and Paraphrase: Some Innovative Uses of Language,

New York: W. W. Norton and Co.Grzega, Joachim (2002), “Some Aspects of Modern Diachronic Onomasiology”. Linguistics 40: 1021-1045.Halle, Morris (1973), “Prolegomena to a Theory of Word Formation”, Linguistic Inquiry 1: 3-16.Jackendoff, Ray S. (1975), “Morphological and Semantic Regularities in the Lexicon”, Language 51: 639-

671.Kiparsky, Paul (1982), “Lexical Morphology and Phonology”, in: Yang, I. S. (ed.), Linguistics in the

Morning Calm: Selected Papers from SICOL-1981, 3-91, Seoul: Hanshin Publishing Company.Krupa, Viktor / Genzor, Jozef / Drozdík, Ladislav (1983), Jazyky sveta, Bratislava: Obzor Lees, Robert B. (1960), The Grammar of English Nominalizations, Bloomington (IN): Indiana University

Press.Lieber, Rochelle (2004), Morphology and Lexical Semantics, Cambridge: Cambridge University Press.Lieber, Rochelle (2005), “English Word-Formation Processes. Observations, Issues, and Thoughts on Future

Research”, in: Štekauer, Pavol / Lieber, Rochelle (eds.), Handbook of Word-Formation, 375-427,Dordrecht: Springer.

Marchand, Hans (1960), The Categories and Types of Present-Day English Word-Formation, Wiesbaden:Otto Harrassowitz.

Marle, Jaap van (1985), On the Paradigmatic Dimension of Morphological Creativity. Dordrecht: Foris.Marle, Jaap van (1986), “The Domain Hypothesis: The Study of Rival Morphological Processes”, Linguistics

24: 601-627.Mohanan, Karuvannur Puthanveettil (1986), The Theory of Lexical Phonology, Dordrecht: Reidel.Plag, Ingo (1996), “Selectional Restrictions in English Suffixation Revisited: A Reply to Fabb (1988)”,

Linguistics 34: 769-798.Plag, Ingo (1999), Morphological Productivity: Structural Constraints in English Derivation, Berlin/New

York: Mouton de Gruyter.Plag, Ingo (2004), “Syntactic Category Information and the Semantics of Derivational Morphological Rules”,

Folia Linguistica 38: 193-225.Raffelsiefen, Renate E. (1992), “A Nonconfigurational Approach to Morphology”, in: Aronoff, Mark (ed.),

Morphology Now, 133-162, Albany: State University of New York Press.Rainer, Franz (1988), “Towards a Theory of Blocking”, in: Booij, Geert / Marle Jaap, van (eds.), Yearbook of

Morphology 1988, 155-185, Dordrecht: Foris.Rainer, Franz (2005), “Constraints on Productivity”, in: Štekauer, Pavol / Lieber, Rochelle (eds.), Handbook

of Word-formation, 335-352, Dordrecht: Springer.Roeper, Thomas and Siegel, M. E. A. (1978), “A Lexical Transformation for Verbal Compounds”, Linguistic

Inquiry 9: 199-260.Scalise, Sergio (1984), Generative Morphology, Dordrecht: Foris.Selkirk, Elisabeth O. (1982), The Syntax of Words, [Linguistic Inquiry Monograph 7], Cambridge (MA): MIT

Press.Štekauer, Pavol (1994), “On Productivity in Word-Formation”, Linguistica Pragensia 2: 67-82.Štekauer, Pavol (1998), An Onomasiological Theory of English Word-Formation, Amsterdam/Philadelphia:

John Benjamins.Štekauer, Pavol (1999), “K teórii produktivity v anglickej slovotvorbe”, Časopis pro moderní filologii 2: 68-

82Štekauer, Pavol (2001). “Fundamental Principles of an Onomasiological Theory of English Word-

Formation”, Onomasiology Online 2: 1-42 [www.onomasiology.de].Štekauer, Pavol (2003), “On Some Issues of an Onomasiological Approach to Productivity”, Ranam 36.

[ESSE 6 – Strasbourg 2002, 2 – linguistics]: 93-99.Štekauer, Pavol (2005a), Meaning Predictability in Word Formation, Amsterdam/Philadelphia: John

Benjamins.Štekauer, Pavol (2005b), “Onomasiological Approach to Word-formation”, in: Štekauer, Pavol / Lieber,

Rochelle (eds.), Handbook of Word-Formation, 207-232, Dordrecht: Springer.

49

Strauss, Steven L. (1982), Lexicalist Phonology of English and German, Dordrecht/Cinnaminson: ForisPublications.

Williams, Edwin (1981), “On the Notions ‘Lexically Related’ and ‘Head of a Word’”, Linguistic Inquiry 2:245-274.

Zimmer, Karl E. (1964), Affixal Negation in English and Other Languages: An Investigation of RestrictedProductivity, [Supplement to Word, Monograph 5], New York.

first version received 24 Januar 2005revised version received 17 July 2005

50Appendix

Word Choices Survey

We are trying to learn more about the words people use for new or unusual situations. We would appreciate your takinga few minutes to fill out this questionnaire. Your participation is entirely voluntary.

INSTRUCTIONS: This questionnaire asks your opinions about words. This is NOT a test, and there are no ”right”answers. We don’t care whether you make up new words for the answers or whether you choose words that already existin English. We just want to see what words you think will work best for a few situations.

Task 1. Choose the word that you think is the most suitable for the person described in the question.

1. A person whose smiling face is used for billboard advertisements:

a. smilerb. smilistc. smilant d. smileman

e. smile-personf. smileg. other: ____________________________

2. A person who dials telephone numbers with a feather:

a. feathererb. featheristc. featherant

d. feathermane. other: _______________________

3. A person who frequently interrupts other people when they are talking: a. interrupter b. interruptist

c. interruptantd. butt-ine. butter-in

f. butter-inner g. butt-innnisth. butt-insky

i. cut-inj. cutter-ink. cutter-innerl. cutt-innistm. cutman

n. interposero. interposistp. other: ______________

4. A person who believes in miracles: a. miraclist or miraculistb. miracler c. miraclant or miraculantd. miracle-man

e. miracle-believerf. miracle-hoperg.miracle-hopisth. miracle-hopei. other: ___________________________

5. A person who is obsessed by something:a. an obsesseeb. an obsessorc. an obsessantd. an obsessist

e. an obsessf. an obsession-mang. an obsessiveh. other: _______________________

51

Task 2. Each question describes a person in an unusual situation. If you had to come up with a name or title for theperson, what would it be? You may make up a word or choose a word that already exists in English.

1. Suppose that space aliens were about to land on Earth for the first time. What would you call a person who wassupposed to meet them as a representative of the human race?

2. What would you call someone who does research about spider webs?

3. What name or title would you use for someone who always tells blond jokes?

4. What name or title would you give a basketball player who always hangs onto the rim after a slam-dunk?

5. Suppose that a woman has a clone made of herself. Then suppose that a man has a clone made of himself. Nowsuppose that the two clones marry each other and have a child. What would you call the child?

Task 3. Each picture below shows a person performing an unusual action. If you had to come up with a name or title forthe person in each picture, what would it be? You may invent a word or choose a word that already exists in English.

1)

What name or title would you give to this person? _______________________

2)

What name or title would you give to this person? ___________________________

3

52

What name or title would you give to this person? _________________________________

4)

What name or title would you give this person? ________________________________

5)

What name or title would you give to this person? ____________________________________

Task 4. In this task, there are five words. Use each word in a sentence, even if you think it isn’t an English word. Thenrate how likely you and other English speakers would be to use the word.

1. engroupment a. Use this word in a sentence:

53

b. How likely would you use this word? Circle one.Extremely Somewhat Likely Very ExtremelyUnlikely Unlikely Likely Likely

c. How likely would others use this word? Circle one.Extremely Somewhat Likely Very ExtremelyUnlikely Unlikely Likely Likely

2. thinnen a. Use this word in a sentence:

b. How likely would you use this word? Circle one.Extremely Somewhat Likely Very ExtremelyUnlikely Unlikely Likely Likely

c. How likely would others use this word? Circle one.Extremely Somewhat Likely Very ExtremelyUnlikely Unlikely Likely Likely

3. swimmeea. Use this word in a sentence:

b. How likely would you use this word? Circle one.Extremely Somewhat Likely Very ExtremelyUnlikely Unlikely Likely Likely

c. How likely would others use this word? Circle one.Extremely Somewhat Likely Very ExtremelyUnlikely Unlikely Likely Likely

4. sleepablea. Use this word in a sentence:

b. How likely would you use this word? Circle one.Extremely Somewhat Likely Very ExtremelyUnlikely Unlikely Likely Likely

c. How likely would others use this word? Circle one.Extremely Somewhat Likely Very ExtremelyUnlikely Unlikely Likely Likely

5. satisfactoritya. Use this word in a sentence:

b. How likely would you use this word? Circle one.Extremely Somewhat Likely Very ExtremelyUnlikely Unlikely Likely Likely

54c. How likely would others use this word? Circle one.

Extremely Somewhat Likely Very ExtremelyUnlikely Unlikely Likely Likely

Demographic Information

This information will be used for statistics only; it won’t be used to identify any individual. You don’t have to finish thisquestionnaire if you don’t want to, but the information is important for our study. If you don’t want to participate, pleasejust keep the questionnaire. If you don’t mind participating, please give answers that are as complete as possible andreturn your questionnaire.

A. PERSONAL INFORMATION

Age: Sex: Where born (state or country):

Occupation:

Spouse’s Occupation:

Father’s Occupation: Where born (state or country):

Mother’s Occupation: Where born (state or country):

Your Education (circle highest level that applies): Some High High School Some College College Graduate Graduate School

School Graduate

B. LANGUAGE BACKGROUND

Please list the languages that you speak and rate your ability according to the following scale:A. I am a native speaker. B. I am not a native speaker, but native speakers usually think that I am. (near-native)C. I speak the language fluently, but I have an accent or sometimes say things that do not sound natural to native

speakers. D. I can speak the language fairly well, but sometimes I have to hesitate to think of words or grammatical constructions. E. I know a little bit, but I have a hard time conversing normally in the language.

Language Ability (A, B, C, D, or E)1. English2.3.

C. YOUR FATHER’S LANGUAGE BACKGROUND

Please list the languages that your father speaks and rate his ability according to the following scale:A. He is a native speaker. B. He is not a native speaker, but native speakers usually think that he is. (Near-native)C. He speaks the language fluently, but he has an accent or sometimes says things that do not sound natural to native

speakers. D. He can speak the language fairly well, but sometimes he has to hesitate to think of words or grammatical

constructions. E. He knows a little bit, but he has a hard time conversing normally in the language.

55Language Ability (A, B, C, D, or E)1. English2.3.

D. YOUR MOTHER’S LANGUAGE BACKGROUND

Please list the languages that your mother speaks and rate her ability according to the following scale:A. She is a native speaker.B. She is not a native speaker, but native speakers usually think that she is. (Near-native)C. She speaks the language fluently, but she has an accent or sometimes says things that do not sound natural to native

speakers. D. She can speak the language fairly well, but sometimes she has to hesitate to think of words or grammatical

constructions. E. She knows a little bit, but she has a hard time conversing normally in the language.

Language Ability (A, B, C, D, or E)1. English2.3.

E. LANGUAGES SPOKEN IN YOUR HOME1._______________________

a. always b. frequently (daily or nearly so) c. occasionally d. almost never e. never2._______________________

a. always b. frequently (daily or nearly so) c. occasionally d. almost never e. never3._______________________

a. always b. frequently (daily or nearly so) c. occasionally d. almost never e. never

May we contact you for help in locating other people who might be willing to complete this survey? Y / N

Your contact information (name, address, phone, e-mail):